Power system scheduling method and device based on uncertainty and computer equipment

Abstract

The invention relates to a power system scheduling method and device based on uncertainty and computer equipment. The method comprises the following steps: acquiring real-time operation data of a power system, and determining an operation state of the power system according to the real-time operation data; determining at least one pre-selected scheduling model from at least one pre-trained candidate scheduling model according to the state type of the running state; according to the real-time operation data and the at least one pre-selected scheduling model, determining a matching degree between the operation state and the at least one pre-selected scheduling model; determining a target scheduling model from the at least one pre-selected scheduling model according to the matching degree of the running state and the at least one pre-selected scheduling model and the real-time running data; and on the basis of the real-time operation data, a scheduling strategy for the power system is generated through the target scheduling model, and the power system is scheduled according to the scheduling strategy, so that the pertinence and effectiveness of the scheduling strategy are improved, and the scheduling accuracy of the power system is ensured.

Description

Technical Field

[0001] The present application relates to the technical field of power dispatching, and particularly to a power system dispatching method, device, computer device, computer-readable storage medium, and computer program product based on uncertainty. Background Art

[0002] With the increasing complexity and dynamics of power systems, the operation and dispatching of power systems face more and more uncertain factors, which mainly come from load fluctuations, intermittent fluctuations of renewable energy, and equipment failures. In order to better cope with the uncertainties of power systems, various uncertainty modeling methods need to be introduced into the optimal dispatching of power systems to improve the economy, reliability, and sustainability of power systems.

[0003] Currently, there are already various uncertainty modeling methods for power systems. Different modeling methods have their own advantages and disadvantages in different application scenarios, but the accuracy of dispatching power systems using existing uncertainty modeling methods is relatively low. Summary of the Invention

[0004] Based on this, it is necessary to provide a power system dispatching method, device, computer device, computer-readable storage medium, and computer program product based on uncertainty that can improve the dispatching accuracy of power systems for the above technical problems.

[0005] In a first aspect, the present application provides a power system dispatching method based on uncertainty, the method comprising:

[0006] Obtain real-time operation data of the power system, and determine the operation state of the power system according to the real-time operation data;

[0007] Determine at least one preselected dispatching model from at least one pre-trained candidate dispatching model according to the state type of the operation state; the candidate dispatching model is trained based on the uncertainty parameters of the power system and historical operation data, and the uncertainty parameters are used to characterize the fluctuation conditions of the power system in the operation states corresponding to their respective state types;

[0008] Determine the matching degree between the operation state and each of the at least one preselected dispatching model according to the real-time operation data and the at least one preselected dispatching model;

[0009] Determine a target dispatching model from the at least one preselected dispatching model according to the matching degree between the operation state and each of the at least one preselected dispatching model and the real-time operation data;

[0010] Based on the real-time operation data, a scheduling strategy for the power system is generated through the target scheduling model, and the power system is scheduled according to the scheduling strategy.

[0011] In one embodiment, the determining the matching degree between the operating state and each of the at least one preselected scheduling model according to the real-time operation data and the at least one preselected scheduling model includes:

[0012] For each of the preselected scheduling models, according to the real-time operation data, determine the model performance parameters of the preselected scheduling model targeted; the model performance parameters are used to characterize the matching degree between the preselected scheduling model targeted and the corresponding state type;

[0013] According to the relative importance of the model performance parameters when the scheduling strategy is generated for the preselected scheduling model targeted, determine the weight coefficient of the model performance parameters;

[0014] According to the model performance parameters and the weight coefficients corresponding to the model performance parameters, determine the matching degree between the preselected scheduling model targeted and the operating state.

[0015] In one embodiment, the model performance parameters include at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model targeted; the determining the model performance parameters of the preselected scheduling model targeted according to the real-time operation data includes:

[0016] According to the real-time operation data, determine at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model targeted.

[0017] In one embodiment, the determining at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model targeted according to the real-time operation data includes at least one of the following:

[0018] According to the data scale of the real-time operation data and the model structure of the preselected scheduling model targeted, respectively determine the time complexity of the preselected scheduling model targeted in terms of running time and the space complexity in terms of storage space, and determine the computational complexity according to the time complexity and the space complexity;

[0019] Determine the running prediction data of the preselected scheduling model targeted at the running moment of the real-time operation data, and determine the model accuracy according to the difference between the real-time operation data and the running prediction data;

[0020] Based on the preselected scheduling model, determine the preselected strategies of the power system under different fluctuation conditions according to the real-time operation data, and determine the model stability parameters according to the fluctuation ranges of the respective preselected strategies;

[0021] Based on the real-time operation data, determine the adaptability index of the preselected scheduling model under the operation state, and determine the model adaptability parameters according to the adaptability index and the index weights corresponding to the respective adaptability indexes.

[0022] In one embodiment, the determining the target scheduling model from the at least one preselected scheduling model according to the matching degree of the operation state and the respective at least one preselected scheduling model and the real-time operation data includes:

[0023] Determine the adaptation parameter of the preselected scheduling model targeted to the operation state according to the matching degree of the operation state and the preselected scheduling model targeted to it and the real-time operation data;

[0024] When the adaptation parameter meets the model parameter update condition, iteratively update the model parameters of the preselected scheduling model targeted to it until the end condition is met to obtain the global optimization parameters;

[0025] Determine the target scheduling model from the at least one preselected scheduling model according to the global optimization parameters corresponding to the respective at least one preselected scheduling model.

[0026] In one embodiment, the method further includes:

[0027] Obtain the scheduling execution result of scheduling the power system based on the scheduling strategy, and determine the scheduling difference between the scheduling execution result and the expected scheduling result;

[0028] When the scheduling difference meets the model update condition, determine the operation data of the power system after scheduling execution;

[0029] Determine the real-time operation data of the power system according to the operation data after scheduling execution, and return to the step of obtaining the real-time operation data of the power system and determining the operation state of the power system according to the real-time operation data, or return to the step of determining the matching degree of the operation state and the respective at least one preselected scheduling model according to the real-time operation data and the at least one preselected scheduling model.

[0030] In a second aspect, the present application further provides a power system scheduling device based on uncertainty, and the device includes:

[0031] A status determination module, configured to obtain real-time operation data of a power system and determine an operation status of the power system according to the real-time operation data;

[0032] A model screening model, configured to determine at least one preselected scheduling model from at least one pre-trained candidate scheduling model according to a status type of the operation status; the candidate scheduling model is trained based on uncertainty parameters and historical operation data of the power system, and the uncertainty parameters are used to characterize fluctuations of the power system in corresponding operation statuses of respective status types;

[0033] A matching degree determination module, configured to determine matching degrees of the operation status and the at least one preselected scheduling model according to the real-time operation data and the at least one preselected scheduling model;

[0034] A model matching module, configured to determine a target scheduling model from the at least one preselected scheduling model according to the matching degrees of the operation status and the at least one preselected scheduling model and the real-time operation data;

[0035] An operation scheduling module, configured to generate a scheduling strategy for the power system through the target scheduling model based on the real-time operation data, and schedule the power system according to the scheduling strategy.

[0036] In a third aspect, the present application further provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the method described above are implemented.

[0037] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described above are implemented.

[0038] In a fifth aspect, the present application further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the method described above are implemented.

[0039] The above-mentioned power system scheduling method, device, computer equipment, computer-readable storage medium and computer program product based on uncertainty obtain real-time operation data of the power system, and determine the operation state of the power system according to the real-time operation data; determine at least one preselected scheduling model from at least one pre-trained candidate scheduling model according to the state type of the operation state; the candidate scheduling model is trained based on the uncertainty parameters and historical operation data of the power system, and the uncertainty parameters are used to characterize the fluctuation of the power system in the operation state corresponding to each state type; determine the matching degree between the operation state and each of the at least one preselected scheduling model according to the real-time operation data and the at least one preselected scheduling model; determine the target scheduling model from the at least one preselected scheduling model according to the matching degree between the operation state and each of the at least one preselected scheduling model and the real-time operation data; generate a scheduling strategy for the power system through the target scheduling model based on the real-time operation data, and schedule the power system according to the scheduling strategy; dynamically determine the operation state of the power system based on the real-time operation data, dynamically evaluate the matching degree between the preselected scheduling model and the current operation state of the power system, select the target scheduling model that matches the current operation state from the preselected scheduling models, and finally generate the corresponding scheduling strategy through the target scheduling model to schedule the power system, which can ensure that the selected target scheduling model accurately adapts to the actual operation state of the power system, dynamically generate the corresponding scheduling strategy according to the real-time operation data, can effectively cope with uncertainty factors such as load fluctuations and renewable energy intermittency, thereby improving the pertinence and effectiveness of the scheduling strategy and ensuring the scheduling accuracy of the power system. Description of the Drawings

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0041] Figure 1 It is an application environment diagram of the power system scheduling method based on uncertainty in an embodiment;

[0042] Figure 2 It is a flowchart of the power system scheduling method based on uncertainty in an embodiment;

[0043] Figure 3 It is a flowchart of the step of determining the matching degree in an embodiment;

[0044] Figure 4 It is a structural block diagram of the power system scheduling device based on uncertainty in an embodiment;

[0045] Figure 5 It is the internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0046] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0047] The power system scheduling method based on uncertainty provided by the embodiments of the present application can be applied to an application environment as Figure 1 shown. Among them, the terminal 102 communicates with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed in the cloud or on other network servers. The server 104 obtains the real-time operation data of the power system and determines the operation state of the power system according to the real-time operation data; the server 104 then determines at least one preselected scheduling model from at least one pre-trained candidate scheduling model according to the state type of the operation state; the candidate scheduling model is trained based on the uncertainty parameters and historical operation data of the power system, and the uncertainty parameters are used to characterize the fluctuation of the power system under the operation states corresponding to their respective state types; subsequently, the server 104 determines the matching degree between the operation state and each of the at least one preselected scheduling model according to the real-time operation data and the at least one preselected scheduling model, and determines the target scheduling model from the at least one preselected scheduling model according to the matching degree between the operation state and each of the at least one preselected scheduling model and the real-time operation data; finally, the server 104 generates a scheduling strategy for the power system through the target scheduling model based on the real-time operation data, and schedules the power system according to the scheduling strategy.

[0048] Among them, the terminal 102 can be but is not limited to various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server 104 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0049] In an exemplary embodiment, as Figure 2As shown, a power system scheduling method based on uncertainty is provided. Taking the method applied to the server 104 in Figure 1 as an example for illustration, it can be understood that the method can also be applied to the terminal 102 in Figure 1 and can also be applied to a system including the terminal 102 and the server 104, which is realized through the interaction between the terminal 102 and the server 104. The method of this embodiment includes the following steps 202 to step 210. Among them:

[0050] Step 202, obtain the real-time operation data of the power system, and determine the operation state of the power system according to the real-time operation data.

[0051] Among them, the real-time operation data refers to the data that can reflect the actual operation status of the power system collected from the power system at the current moment; the real-time operation parameters can include, but are not limited to, electrical parameters, equipment status parameters, load parameters, renewable energy parameters, etc. For example, the electrical parameters can include voltage, current, frequency, power (active power, reactive power), etc.; the equipment status parameters can include the status parameters of equipment such as generators, transformers, and transmission lines, such as the switch status (on or off), temperature, etc. of the equipment; the load parameters can include the electricity consumption demand data of each user in the power system, such as the load size and load distribution in different time periods; the renewable energy parameters can include the power generation of renewable energy participating in power generation in the power system, etc.

[0052] The operation state refers to the overall working condition of the power system determined according to the real-time operation data at the current moment. The operation state can include normal operation state, abnormal operation state, and fault operation state, etc. For example, the normal operation state means that all parameters in the power system are within the normal range, the equipment operates stably, and it can meet the electricity consumption needs of users. At this time, parameters such as voltage and frequency are maintained near the rated value, and there is no abnormal alarm for power generation and transmission equipment; the abnormal operation state means that there are phenomena deviating from the normal situation in the system, but it has not reached the level of a fault, such as slight overload of some lines, or the temperature of some equipment is slightly higher than the normal range, but still within the acceptable safety range; the fault operation state means that obvious faults occur in the power system, such as line short circuits, equipment damage, etc., resulting in serious deviation of system parameters from the normal values and the system cannot operate normally.

[0053] Exemplarily, the server can obtain the real-time operation data of the power system from the data acquisition devices on each device in the power system. The data acquisition devices can include various current transformers and voltage transformers for collecting current and voltage, transmitters for collecting power, electric meters for collecting current, frequency meters for collecting frequency, and various detection devices for collecting the status data of power generation equipment, substation equipment, and transmission lines, etc. The server can also determine the operation state of the power system according to whether the real-time operation data is within the normal range.

[0054] Step 204: Determine at least one preselected scheduling model from at least one pre-trained candidate scheduling model according to the state type of the operating state.

[0055] Among them, the candidate scheduling model refers to a model trained based on the uncertainty parameters and historical operation data of the power system, which can be used to generate scheduling strategies for the power system under different operating states. The uncertainty parameters are used to characterize the fluctuations of the power system under the operating states corresponding to their respective state types. For example, in load forecasting, due to the randomness and uncertainty of users' electricity consumption behaviors, there will be a certain range of fluctuations in load data. At this time, the fluctuations in load data can be described as uncertainty parameters. Another example is in the forecasting of renewable energy generation (such as wind power generation, photovoltaic power generation, etc.), since the generation of renewable energy usually directly depends on the fluctuations of natural conditions (such as wind speed, light intensity, cloud cover, etc.), the generated electricity of renewable energy has uncertainty. At this time, the generation fluctuations of renewable energy can also be described as uncertainty parameters. The uncertainty parameters can help the model better adapt to the changes in the operation of the power system and improve the robustness of the scheduling strategy. The historical operation data can include various information of the power system in the past under different operating states, such as load curves, unit output / start-stop, generation plans, transmission schemes, load changes, electrical parameters, etc. Specifically, in implementation, the historical operation data can be obtained through the SCADA system (Supervisory Control And Data Acquisition) of the power system, so that the candidate scheduling model can determine the scheduling strategies under different operating states through learning the historical operation data, thereby providing a reference for the current power dispatching decision.

[0056] The preselected scheduling model refers to a model selected from at least one candidate scheduling model according to the state type of the operating state of the power system and applicable to the current operating state. The state type of the operating state can include normal, abnormal, faulty, etc. Among normal and abnormal, it can also include various uncertainties in the power system. For example, when the power system is in a normal operating state, those candidate scheduling models that perform well under normal load conditions can be selected as preselected scheduling models. When the system is in an abnormal or faulty operating state, models that can handle the corresponding abnormal or faulty situations can be selected. Another example is that in an operating state with large load fluctuations, a model that can handle load fluctuations can be selected, and in an operating state with large fluctuations in renewable energy generation, a model that can handle renewable energy fluctuations can be selected.

[0057] In some exemplary embodiments, the candidate scheduling model can be constructed by adopting at least one of a probability modeling method (such as Monte Carlo simulation, probabilistic constraint optimization), a fuzzy set theory modeling method (such as fuzzy logic control), a rough set theory modeling method, an artificial intelligence modeling method (such as neural network, genetic algorithm, particle swarm optimization, etc.). The candidate scheduling models obtained by using different modeling methods can handle the scheduling decisions of the power system in different operating states.

[0058] The probability modeling method can describe the uncertainty in the power system by introducing random variables. For example, in the power system, if a certain load or the power generation of renewable energy is a random variable, the goal of the probability modeling method is to solve the optimal scheduling scheme, that is:

[0059] (1) [[ID=...]] [[ID=...]]

[0060] where is the scheduling cost function; is the probability density function of the uncertainty variable; is the uncertainty space; by randomly sampling to simulate multiple operating scenarios, the scheduling cost of each scenario can be determined, and then the overall expected cost can be obtained through statistical analysis.

[0061] The fuzzy set theory modeling method can be used to handle situations with fuzziness and uncertainty. For example, assuming that the load demand in the power system is a fuzzy set, then the fuzzy rule can be defined for reasoning:

[0062] (2)

[0063] where is the fuzzy rule; is the fuzzy linguistic variable of the load demand , such as "high load", "low load", etc.; is the control variable, such as the output power of the generator; [[ID=...]] is the control variable 's fuzzy linguistic variable; that is, based on the fuzzy rule , when belongs to the set , then belongs to the set . In this way, after defuzzification through fuzzy reasoning, a specific scheduling scheme can be obtained.

[0064] The rough set theory can be used to handle problems of incomplete information. For example, in the power system, assuming that the state set of the power system is incomplete or fuzzy, is the state of the power system, is the number of states. When using rough set theory for uncertainty modeling, the goal of rough set theory can be used for decision support through the approximation of concepts, that is:

[0065] (3)

[0066] Among them, is the equivalence relation between state concepts. By approximating classification and reduction to handle the uncertainty in data, it can provide a decision for a definite scheduling scheme.

[0067] Artificial intelligence modeling methods can be used for load forecasting or power generation forecasting. For example, in a power system, when using a neural network for load forecasting or power generation forecasting, assume that the historical data of load and renewable energy power generation is , is the historical data, is the number of historical data. The neural network learns the non-linear relationship between historical data to predict the future load or the power generation of renewable energy, that is:

[0068] (4)

[0069] Among them, is the load or the power generation of renewable energy predicted at the moment, is the mapping function of the neural network.

[0070] In specific implementation, multiple candidate scheduling models obtained through training can be stored in a pre-created model library. For each operating state of the power system, there is at least one type of candidate scheduling model in the model library that can handle the uncertainty in this operating state. Each type of candidate scheduling model in the model library corresponds to an operating state, and the two can be determined through a defined mapping relationship or label information, so that the preselected scheduling model can be matched from the model library through the determined operating state.

[0071] Exemplarily, the server can determine a candidate scheduling model that can handle the current state type from the candidate scheduling models according to the mapping relationship between the state type of the operating state of the power system and each candidate scheduling model in the model library, so as to obtain the preselected scheduling model.

[0072] In some other embodiments, the operating status of the power system may also be affected by climate or weather factors, which in turn affect power dispatching. For example, under extreme weather conditions (such as heavy rain, typhoons, etc.), the power system needs to adjust the power generation plan and transmission plan according to meteorological data to ensure the stability and security of power supply. Therefore, when determining the pre-selected dispatching model, meteorological data can also be integrated for comprehensive determination, such as setting weights according to the correlation between the status type and meteorological data, or determining the correlation coefficient, so as to match it with the candidate dispatching model in the model library, and then obtain the pre-selected dispatching model.

[0073] Step 206 : Determine the matching degree between the operating state and the at least one pre-selected scheduling model based on the real-time operating data and the at least one pre-selected scheduling model.

[0074] The matching degree is a parameter used to measure the degree of adaptability between the power system's operating state and each preselected dispatch model. The higher the matching degree, the more suitable the preselected dispatch model is for the current operating state. In practice, the matching degree between the operating state and the preselected dispatch model can be scored to obtain the matching degree between the operating state and the preselected dispatch model.

[0075] Exemplarily, the server may score the degree of fit between the operating state and the preselected scheduling model based on the real-time operating data and the corresponding preselected scheduling model, and determine the degree of fit between the operating state and at least one preselected scheduling model based on the score.

[0076] Step 208 : determining a target scheduling model from the at least one pre-selected scheduling model according to the matching degree between the operation state and the at least one pre-selected scheduling model and the real-time operation data.

[0077] Among them, the target scheduling model refers to a scheduling model suitable for the current power system operating state determined from the pre-selected scheduling model based on the matching degree between the operating state and at least one pre-selected scheduling model and real-time operating data. The determined target scheduling model can effectively meet the current power system operating requirements and optimize system performance (such as improving power supply reliability, reducing power generation costs, etc.); for example, when real-time operating data shows that the system load fluctuates greatly, a pre-selected scheduling model with strong adaptability to load changes may be selected as the target scheduling model.

[0078] For example, the server may employ an optimization algorithm to solve and optimize the matching degree between the operating state and at least one pre-selected scheduling model and the real-time operating data, thereby determining a target scheduling model from the at least one pre-selected scheduling model. The optimization algorithm may employ one or more methods including, but not limited to, particle swarm optimization, ant colony algorithm, genetic algorithm, and gradient descent.

[0079] Step 210: Based on the real-time operation data, generate a scheduling strategy for the power system through the target scheduling model, and schedule the power system according to the scheduling strategy.

[0080] Among them, the scheduling strategy refers to the specific scheduling plan for the power system. The scheduling strategy is generated by the target scheduling model based on the real-time operation data, and the scheduling strategy may include, but is not limited to, generation plans, transmission schemes, reserve arrangements, etc. The generation plan can be used to determine the power generation output of each power plant, including the start-stop arrangement of generators, the allocation of power generation power, etc., to meet the load demand of the system. The transmission scheme can be used to plan the power transmission path and transmission capacity to ensure that electric energy can be safely and efficiently transmitted from the power plant to the user side. The reserve arrangement can be used to determine the reserve capacity of the system to cope with possible emergencies, such as generator failures, sudden load increases, etc.

[0081] Scheduling the power system means controlling and regulating the power generation, transmission, distribution and other links in the power system according to the scheduling strategy to achieve the safe, stable and economic operation of the power system. During specific scheduling, it can be that the server generates corresponding control instructions according to the scheduling strategy to control the equipment in the power system through the automation equipment and control system in the power system, so as to realize the regulation of the output of the power generation equipment, the switching operation of the transmission line, etc.; it can also be that the dispatcher makes manual decisions and operations according to the scheduling strategy and in combination with the actual situation.

[0082] Exemplarily, the server can input the real-time operation data into the target scheduling model, and generate a scheduling strategy for the power system through the target scheduling model based on the real-time operation data. The server can also generate control instructions for each power equipment in the power system according to the scheduling strategy, and apply the control instructions to the power equipment in the power system to control the power equipment to operate according to the control instructions, so as to realize the scheduling of the power system.

[0083] In the above power system scheduling method based on uncertainty, the real-time operation data of the power system is obtained, and the operation state of the power system is determined according to the real-time operation data; at least one preselected scheduling model is determined from at least one pre-trained candidate scheduling model according to the state type of the operation state; the candidate scheduling model is trained based on the uncertainty parameters and historical operation data of the power system, and the uncertainty parameters are used to characterize the fluctuation of the power system under the operation states corresponding to their respective state types; according to the real-time operation data and at least one preselected scheduling model, the matching degrees of the operation state and each of the at least one preselected scheduling models are determined; according to the matching degrees of the operation state and each of the at least one preselected scheduling models and the real-time operation data, a target scheduling model is determined from the at least one preselected scheduling models; based on the real-time operation data, a scheduling strategy for the power system is generated through the target scheduling model, and the power system is scheduled according to the scheduling strategy; the operation state of the power system is dynamically determined based on the real-time operation data, and the matching degree between the preselected scheduling model and the current operation state of the power system is dynamically evaluated. According to the matching degree, a target scheduling model that matches the current operation state is selected from the preselected scheduling models. Finally, a corresponding scheduling strategy is generated through the target scheduling model to schedule the power system, which can ensure that the selected target scheduling model accurately adapts to the actual operation state of the power system, and a corresponding scheduling strategy is dynamically generated according to the real-time operation data, which can effectively cope with uncertainty factors such as load fluctuations and intermittent renewable energy, thereby improving the pertinence and effectiveness of the scheduling strategy and ensuring the scheduling accuracy of the power system.

[0084] In some embodiments, as Figure 3 shown, step 206 includes the following steps 302 to 306. Among them:

[0085] Step 302, for each preselected scheduling model, according to the real-time operation data, determine the model performance parameters of the preselected scheduling model being targeted.

[0086] Among them, the model performance parameter refers to an index used to quantitatively describe the matching degree between the preselected scheduling model and the corresponding state type under the real-time operation data, so as to reflect the performance of the preselected scheduling model in terms of accuracy, stability, adaptability, etc. The model performance parameter can be used as an important basis for evaluating the pros and cons of the preselected scheduling model to further select a more suitable scheduling model for the current operation state from at least one preselected scheduling model.

[0087] Exemplarily, the server can, for each preselected scheduling model, according to the real-time operation data, and use different methods to determine the index of the matching degree between the preselected scheduling model being targeted and the corresponding state type, and quantify each index to determine the model performance parameters of the preselected scheduling model being targeted.

[0088] Step 304: Determine the weight coefficient of the model performance parameters according to the relative importance of the model performance parameters when the scheduling policy is generated for the targeted preselected scheduling model.

[0089] Among them, the relative importance refers to the relative magnitude of the influence of each model performance parameter on the scheduling policy generation result when evaluating the preselected scheduling model. Different model performance parameters may have different importance under different operating states and requirements. The relative importance can be determined through expert experience, historical data analysis, experimental verification, etc. For example, in some cases, the accuracy of the scheduling policy may be more important than adaptability; while in other cases, stability may be the primary consideration factor. Therefore, assigning different weight coefficients to different model performance parameters can increase the accuracy of the matching degree determination.

[0090] The weight coefficient refers to a value assigned to each model performance parameter according to the relative importance of the model performance parameters, which is used to quantify the role of the model performance parameter in the determination of the matching degree. The magnitude of the weight coefficient can reflect the proportion of the model performance parameter in the overall determination of the matching degree, so as to adjust the influence degree of each performance parameter when comprehensively calculating the matching degree between the preselected scheduling model and the operating state, and make the calculation result of the matching degree more accurately reflect the actual performance of the model. In specific implementation, the weight coefficient satisfies non-negativity, and the sum of the weight coefficients of all model performance parameters is usually 1 to ensure the rationality and comparability of the evaluation results.

[0091] Exemplarily, the server can determine the relative importance of each model performance parameter when the scheduling policy is generated for the targeted preselected scheduling model through the analysis of historical data, and then determine the weight coefficient of each model performance parameter according to the respective relative importance of each model performance parameter.

[0092] Step 306: Determine the matching degree between the targeted preselected scheduling model and the operating state according to the model performance parameters and the weight coefficients corresponding to the model performance parameters.

[0093] Among them, the matching degree can comprehensively consider the model performance parameters and their corresponding weight coefficients to quantitatively evaluate the adaptation degree between the preselected scheduling model and the current operating state of the power system. The higher the matching degree, the more suitable the preselected scheduling model is for the current operating state. When determining the matching degree based on each model performance parameter and the weight coefficient, the weighted average method is usually adopted, that is, multiplying the value of each model performance parameter by its corresponding weight coefficient, and then adding all the products to obtain the value of the matching degree. As the basis for selecting the final scheduling model, the matching degree enables the server to select a model suitable for the current operating state from multiple preselected scheduling models to generate a scheduling policy that matches the current operating state of the power system.

[0094] Exemplarily, the server may adopt the weighted average method, multiply the value of each model performance parameter by its corresponding weight coefficient, and then sum all the products to determine the matching degree of the preselected scheduling model for the operating state.

[0095] In some alternative embodiments, the model performance parameters include at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model.

[0096] Among them, the computational complexity refers to the computational resources and time required for the preselected scheduling model to process real-time operation data. The computational complexity is usually related to the algorithm structure of the model, the data scale, and the complexity of the computational steps. As an index to measure the processing efficiency of the preselected scheduling model for real-time operation data, in power system scheduling, it is necessary to quickly respond to system changes. Therefore, the lower the computational complexity of the preselected scheduling model, the more it can meet the requirements of real-time scheduling. Specifically, when implementing, the computational complexity of the preselected scheduling model can be evaluated by analyzing the algorithm complexity of the model (such as time complexity, space complexity) or indicators such as the actual computational time and memory occupancy during operation.

[0097] The model accuracy refers to the degree of coincidence between the scheduling strategy generated by the preselected scheduling model and the actual operation requirements of the power system, and is used to reflect the prediction ability of the preselected scheduling model for the operating state of the power system and the effectiveness of the scheduling strategy. As an index to measure the accuracy of the model, a high-precision model can generate a scheduling strategy closer to the actual requirements, thereby improving the operating efficiency and stability of the power system. Specifically, when implementing, the model accuracy can be evaluated by comparing the scheduling strategy generated by the model with the actual operation data (such as load prediction error, deviation between generation plan and actual generation).

[0098] The model stability parameter refers to the parameter of the ability of the preselected scheduling model to maintain its performance stable in the face of noise, outliers, or system state changes in real-time operation data, and is used to reflect the robustness of the preselected scheduling model to uncertainty. As an index to measure the stability of the preselected scheduling model, since the operation data often has noise and uncertainty, in the power system, a stable preselected scheduling model can better maintain its performance in various situations and avoid large changes in the scheduling strategy due to data fluctuations. Specifically, when implementing, the model stability can be evaluated by analyzing the performance of the model under different noise levels, outliers, or system state changes multiple times (such as the fluctuation range of the prediction error, the stability of the scheduling strategy, etc.).

[0099] The model adaptability parameter refers to the parameter of the preselected scheduling model's ability to adjust its scheduling strategy to adapt to the new state when facing the changes in the operating state of the power system, which is used to reflect the flexibility and adjustability of the preselected scheduling model. The operating state of the power system is constantly changing, so the model adaptability parameter needs to have good adaptability in order to generate effective scheduling strategies under different states. In specific implementation, the model adaptability can be evaluated by analyzing the performance of the model under different operating states (such as the adjustment speed of the scheduling strategy, the degree of adaptation to the new state, etc.).

[0100] Taking the model performance parameters including the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model as examples, the weight coefficients are determined for the computational complexity, model accuracy, model stability parameter, and model adaptability parameter respectively. Then, the computational complexity, model accuracy, model stability parameter, and model adaptability parameter are multiplied by their respective corresponding weight coefficients, and all the products are added together to obtain the value of the matching degree.

[0101] In an exemplary embodiment, the matching degree can be expressed as:

[0102] (5)

[0103] Wherein, is the matching degree of the th preselected scheduling model and the operating state; are respectively the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the th preselected scheduling model; are respectively the weight coefficients corresponding to the computational complexity, model accuracy, model stability parameter, and model adaptability parameter.

[0104] In this embodiment, by determining the model performance parameters of the preselected scheduling model, assigning weight coefficients to each model performance parameter according to the relative importance of the model performance parameters when the preselected scheduling model executes the scheduling strategy generation, and determining the matching degree based on the model performance parameters and their corresponding weight coefficients, it is possible to scientifically and reasonably quantify the matching degree between the preselected scheduling model and the operating state, which is beneficial to accurately and quickly select the scheduling model suitable for the current operating state in the subsequent steps.

[0105] In some embodiments, according to the real-time operation data, the model performance parameters of the preselected scheduling model are determined, including:

[0106] According to the real-time operation data, at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model is determined.

[0107] Exemplarily, taking the model performance parameters including computational complexity, model accuracy, model stability parameters, and model adaptability parameters as an example, the server can input the real-time operation data into each of the preselected scheduling models targeted, and respectively determine the computational complexity, model accuracy, model stability parameters, and model adaptability parameters of the preselected scheduling models targeted. In specific implementation, the server can calculate the computational complexity of the model by analyzing the algorithm structure of the preselected scheduling model targeted and the computational resource occupancy during actual operation. The server can evaluate the accuracy of the model by comparing the scheduling strategy generated by the preselected scheduling model targeted with the real-time operation data and calculating indicators such as prediction error and deviation. The server can evaluate the stability of the model by running the preselected scheduling model targeted multiple times under different noise levels, outliers, or system state changes and observing its performance, and calculating indicators such as the fluctuation range of the prediction error. The server can evaluate the adaptability of the model by running the preselected scheduling model targeted under different operating states and observing the adjustment speed and adaptability degree of its scheduling strategy.

[0108] In this embodiment, by accurately evaluating the model performance indicators such as the computational complexity, model accuracy, model stability, and model adaptability of the preselected scheduling model targeted through real-time operation data, it is possible to provide a quantitative and objective basis for model performance analysis, which helps to quickly identify the advantages and disadvantages of the preselected scheduling model targeted under different operating states, and ensure that the selected preselected scheduling model can efficiently and accurately adapt to the actual power system operation requirements.

[0109] In some embodiments, determining at least one of the computational complexity, model accuracy, model stability parameters, and model adaptability parameters of the preselected scheduling model targeted according to real-time operation data includes at least one of the following:

[0110] According to the data scale of the real-time operation data and the model structure of the preselected scheduling model targeted, respectively determine the time complexity in terms of running time and the space complexity in terms of storage space of the preselected scheduling model targeted, and determine the computational complexity according to the time complexity and the space complexity.

[0111] Among them, the data scale of the real-time operation data refers to the size or quantity of the real-time operation data, and the data scale can include but is not limited to the number of data points, the dimensions of the data (such as multiple parameters such as voltage, current, frequency, etc.), and the time span of the data. The model structure of the preselected scheduling model refers to the internal organization method and algorithm logic of the preselected scheduling model, and the model structure can include but is not limited to the hierarchical structure of the preselected scheduling model, the number of nodes, the connection method, and the type of algorithm adopted (such as neural network, decision tree, linear regression, etc.).

[0112] The time complexity refers to the time resources required for the pre-selection scheduling model to process real-time operation data, which is used to reflect the trend of the time required for the pre-selection scheduling model to process data as the data scale increases. The time complexity can be expressed as O(n), where n represents the data scale. For example, for a pre-selection scheduling model constructed based on the Monte Carlo simulation method, its time complexity can be expressed as O(N1×T1), where N1 is the sampling times of the real-time operation data, and T1 is the running time of the pre-selection scheduling model; another example is that for a pre-selection scheduling model constructed based on the probability constraint optimization method, its time complexity can be expressed as O(N2 3 ), where N2 is the dimension of the real-time operation data. Also, for a pre-selection scheduling model constructed based on a neural network, its time complexity can be expressed as O(L×N3×M), where L is the number of layers of the neural network, N3 is the number of nodes in each layer, and M is the number of training samples of the model. The space complexity refers to the storage space resources required for the pre-selection scheduling model to process real-time operation data, which is used to reflect the size of the memory or disk space required by the pre-selection scheduling model during data storage and processing. The space complexity can also be expressed as O(n), and for specific details, please refer to the description of the time complexity. The computational complexity can comprehensively consider the time complexity and the space complexity to overall evaluate the computing resources and time required for the pre-selection scheduling model to process real-time operation data.

[0113] Exemplarily, the server can respectively determine the time complexity of the targeted pre-selection scheduling model in terms of running time and the space complexity in terms of storage space according to the data scale of the real-time operation data and the model structure of the targeted pre-selection scheduling model, and determine the computational complexity based on the time complexity and the space complexity. For example, the server can, according to the data dimension of the real-time operation data and the number of model layers, the number of nodes in each layer, etc., of the targeted pre-selection scheduling model, and according to the method selected for the corresponding targeted pre-selection scheduling model, respectively determine the time complexity of each targeted pre-selection scheduling model in terms of time operation and the space complexity in terms of storage space, and comprehensively consider the time complexity and the space complexity to obtain the computational complexity.

[0114] In some alternative embodiments, when determining the computational complexity, the server can, according to the characteristics of the pre-selection scheduling model, respectively assign corresponding weights to the time complexity and the space complexity to perform weighted fusion of the time complexity and the space complexity to obtain the computational complexity. For example, for a pre-selection scheduling model with higher requirements for storage space, a larger weight can be assigned.

[0115] In this embodiment, by combining the data scale of the real-time operation data and the model structure of the pre-selection scheduling model, the time complexity and the space complexity can be accurately determined, and then the computational complexity can be obtained, which is beneficial to providing data support for model decision-making to improve the overall system performance and resource utilization efficiency.

[0116] Determine the operation prediction data at the operation moment of the real-time operation data for the preselected scheduling model, and determine the model accuracy according to the difference between the real-time operation data and the operation prediction data.

[0117] Among them, the operation prediction data refers to the operation state or parameter value of the power system predicted by the preselected scheduling model at the operation moment of the real-time operation data according to the internal logic and algorithm of the model. By comparing the operation prediction data with the actual value of the real-time operation data, the prediction error can be calculated. Specifically, when implemented, the operation prediction data can be the data predicted by the preselected scheduling model based on the data before the operation moment through internal logic and algorithm. For example, the preselected scheduling model predicts based on the operation data at time T-1 to obtain the operation prediction data at time T. When the power system runs to time T, time T is the operation moment of the real-time operation data. At this time, the real-time operation data at time T is compared with the operation prediction data at time T to determine the difference between the real-time operation data and the operation prediction data, and then determine the model accuracy. The model accuracy can be measured by indicators such as prediction error and accuracy rate, such as mean square error, absolute error, etc. In this way, the model accuracy can be evaluated according to the size of the error.

[0118] Exemplarily, the server can input the operation data of the power system before the operation moment of the real-time operation data into the preselected scheduling model, predict the operation prediction data of the power system at the operation moment through the internal logic and algorithm of the preselected scheduling model. Subsequently, the server can calculate the mean square error between the real-time operation data and the operation prediction data to determine the model accuracy.

[0119] In an optional embodiment, the model accuracy determined by using the mean square error is expressed as:

[0120] (6)

[0121] Among them, is the mean square error; is the th operation prediction data, is the th real-time operation data, , is the number of operation prediction data and real-time operation data.

[0122] In this embodiment, through the method of predicting in advance and comparing later, the operation prediction data and the real-time operation data to be compared can be on the same time dimension to obtain accurate model accuracy.

[0123] Based on the targeted pre-selection scheduling model, determine the pre-selection strategies of the power system under different fluctuation conditions based on real-time operation data, and determine the model stability parameters according to the fluctuation ranges of each pre-selection strategy.

[0124] Among them, the pre-selection strategy refers to the scheduling strategy or scheduling plan generated by the targeted pre-selection scheduling model based on real-time operation data for the power system under different fluctuation conditions, which is used to reflect the adaptability and flexibility of the targeted pre-selection scheduling model under different operating conditions. Different fluctuation conditions can include but are not limited to load fluctuations, fluctuations in renewable energy power generation, voltage fluctuations, etc. The fluctuation range refers to the fluctuation or change range of the pre-selection strategies generated during the operation of the power system under different fluctuation conditions. The smaller the fluctuation range, the better the stability of the targeted pre-selection scheduling model. Thus, the model stability parameters of the targeted pre-selection scheduling model can be obtained. The model stability parameters can be comprehensively evaluated based on the performance of each pre-selection strategy under different fluctuation conditions (such as the adjustment frequency, adjustment amplitude, and contribution to system stability of the strategy). By analyzing the performance of the pre-selection strategy within the fluctuation range, calculate the stability indicators of the pre-selection strategy (such as volatility, stability coefficient, etc.), and determine the model stability parameters according to the stability indicators.

[0125] Exemplarily, the server can simulate the operation of the power system under different fluctuation conditions, and for different fluctuation conditions, input the real-time operation data into the targeted pre-selection scheduling model to generate the pre-selection strategies of the power system under different fluctuation conditions respectively, and determine the model stability parameters by comparing the fluctuation ranges between different pre-selection strategies.

[0126] In this embodiment, by processing the real-time operation data multiple times under different fluctuation conditions through the pre-selection scheduling model, multiple pre-selection strategies can be obtained, and then the stability parameters are determined based on the fluctuation ranges of the pre-selection strategies, which is beneficial to quantitatively evaluate the stability of the pre-selection scheduling model.

[0127] Based on the real-time operation data, determine the adaptability indicators of the targeted pre-selection scheduling model in the operating state, and determine the model adaptability parameters according to the adaptability indicators and the corresponding index weights of each adaptability indicator.

[0128] Among them, the adaptability index refers to various indexes for evaluating the adaptability and flexibility of a preselected scheduling model in the running state based on real-time operation data, so as to reflect the adaptability and adjustability of the preselected scheduling model under different operating conditions. The adaptability index may include but is not limited to accuracy, tracking performance, robustness, response speed, etc. Accuracy refers to the prediction error of the preselected scheduling model for data within a known time window. Tracking performance refers to the real-time adjustment ability of the preselected scheduling model when the input fluctuates greatly (such as the automatic update ability of model parameters). Robustness refers to the anti-disturbance ability of the preselected scheduling model against abnormal data or noise input. Response speed refers to whether the running time of the preselected scheduling model meets the real-time requirement. The index weight refers to the proportion of each adaptability index in comprehensively evaluating the adaptability of the model, and is used to reflect the importance of each adaptability index in comprehensively evaluating the adaptability of the model; the index weight can be determined according to the requirements of the actual application scenario and the expert's experience judgment. The model adaptability parameter can be obtained by weighted average or other comprehensive evaluation methods according to the adaptability index and its corresponding index weight. For example, multiply the value of each adaptability index by its corresponding index weight, and then add up all the products to obtain the value of the model adaptability parameter. The higher the value of the model adaptability parameter, the stronger the adaptability of the preselected scheduling model to the state change of the power system.

[0129] Exemplarily, the server may input the real-time operation data into the preselected scheduling model targeted, so as to determine the parameters corresponding to the accuracy, tracking performance, robustness, and response speed of the preselected scheduling model targeted in the running state, multiply the parameters of accuracy, tracking performance, robustness, and response speed by their respective corresponding index weights, and then add up all the products to determine the model adaptability parameter.

[0130] In this embodiment, determining the adaptability index of the preselected scheduling model based on the real-time operation data and calculating the adaptability parameter in combination with the index weight are beneficial to accurately evaluating the adaptability parameter of the model in the running state.

[0131] In some embodiments, step 208 includes:

[0132] Determine the adaptation parameter of the preselected scheduling model targeted with respect to the running state according to the matching degree between the running state and the preselected scheduling model targeted and the real-time operation data; when the adaptation parameter meets the model parameter update condition, iteratively update the model parameters of the preselected scheduling model targeted until the end condition is met to obtain the global optimization parameter; determine the target scheduling model from at least one preselected scheduling model according to the global optimization parameters corresponding to each of the at least one preselected scheduling model.

[0133] Among them, the adaptation parameter refers to a parameter used to describe the degree of adaptation between the targeted preselected scheduling model and the operating state under specific operating conditions and real-time operating data, and is used to reflect the adaptability and performance of the preselected scheduling model under the current operating conditions. The model parameter update condition refers to the condition that needs to be satisfied before iteratively updating the model parameters; for example, the adaptation parameter does not reach the preset threshold or target value; the performance of the model under real-time operating data does not meet the expected requirements; the operating state has changed significantly, and it is necessary to adjust the model parameters to adapt to the new operating conditions, etc. The global optimization parameter refers to the parameter obtained after iteratively updating the model parameters of the preselected scheduling model until the end condition is met. The global optimization parameter can enable the preselected scheduling model to achieve the optimal or near-optimal performance under the given operating state. When the adaptation parameter meets the model parameter update condition, the model parameters of the preselected scheduling model are iteratively updated. After each update, the performance of the preselected scheduling model is re-evaluated, and it is determined whether to continue the update based on the new adaptation parameter. The iterative process continues until the end condition is met (such as reaching the preset number of iterations, convergence of the adaptation parameter, etc.).

[0134] Exemplarily, taking the particle swarm optimization method as an example, the server can randomly generate multiple particles according to each preselected scheduling model and the corresponding initial model parameters of each preselected scheduling model. Each particle represents a combination of a preselected scheduling model and its corresponding initial model parameters, and the position and velocity of each particle are initialized; for each particle, the server can calculate the adaptation parameter of each particle according to the fitness function. For example, the server can comprehensively determine the adaptation parameter of the preselected scheduling model according to the performance metrics (such as accuracy, precision, F1 score, etc.) and business metrics (such as real-time performance, resource consumption, etc.) of each preselected scheduling model. For example, the adaptation parameter can be determined by weighted summation of the performance metrics and business metrics; then, the server can compare the adaptation parameter with the individual optimal solution and the global optimal solution in the particle swarm to determine whether the adaptation parameter meets the model parameter update condition. For example, when the adaptation parameter is better than the individual optimal solution, the server can update the particle using the individual optimal solution. When the adaptation parameter is greater than the global optimal solution, the server can update the particle using the global optimal solution and update the position and velocity of the particle according to the update formula until the maximum number of iterations is reached or the adaptation parameter converges, and the global optimal solution at this time is obtained; finally, the server can determine the corresponding preselected scheduling model and its corresponding model parameter combination as the target scheduling model according to the global optimal solution after the adaptation parameter converges.

[0135] In an alternative embodiment, the update formula of the particle swarm optimization is expressed as:

[0136] (7)

[0137] (8)

[0138] Wherein, are respectively the position and velocity of the th particle (i.e., the th preselected scheduling model) in the th update round; are respectively the position and velocity of the th particle (i.e., the th preselected scheduling model) in the th update round; is the inertia weight; are respectively the individual optimal solution and the global optimal solution; are respectively the individual and global learning factors; are respectively the individual and global random factors.

[0139] In this embodiment, for each preselected scheduling model, its global optimization parameters are obtained through iterative update, the global optimization parameters of each preselected scheduling model and their corresponding performance performances (such as model accuracy, stability, adaptability, etc.) are compared, and according to the comparison results, the preselected scheduling model with the optimal performance or meeting specific requirements is selected as the target scheduling model, which can provide an optimal or near-optimal scheduling strategy under a given operating state, thereby improving the operating efficiency and stability of the power system.

[0140] In some embodiments, the power system scheduling method based on uncertainty further includes:

[0141] Obtain the scheduling execution result of scheduling the power system based on the scheduling strategy, and determine the scheduling difference between the scheduling execution result and the expected scheduling result; when the scheduling difference meets the model update condition, determine the operating data of the power system after scheduling execution; determine the real-time operating data of the power system according to the operating data after scheduling execution, and return to the step of obtaining the real-time operating data of the power system and determining the operating state of the power system according to the real-time operating data, or return to the step of determining the matching degree between the operating state and each of at least one preselected scheduling model according to the real-time operating data and at least one preselected scheduling model.

[0142] Among them, the dispatching execution result refers to the actual operating state and parameter changes presented by the power system after actual dispatching based on the dispatching strategy. The dispatching execution result can include, but is not limited to, the actual values of parameters such as voltage, current, frequency, and power of the power system after dispatching execution, as well as the operating state of power equipment, load distribution, etc. The dispatching execution result can be obtained in real time through the monitoring system and data acquisition device of the power system. The expected dispatching result refers to the operating state and parameter changes that are expected to be achieved by the power system when generating the dispatching strategy, based on factors such as the operating state of the power system, load forecasting, and power generation plan. The expected dispatching result is the goal and desired result of the dispatching strategy. The dispatching difference refers to the difference between the dispatching execution result and the expected dispatching result, which is used to reflect the deviation and uncertainty of the dispatching strategy during the actual execution process. The dispatching difference can be determined by calculating the deviation between the dispatching execution result and the expected dispatching result. The dispatching difference can include, but is not limited to, differences in parameter values, differences in state changes, etc. By comparing the dispatching execution result with the expected dispatching result, the execution deviation of the dispatching strategy can be quantified to analyze the execution effect of the dispatching strategy, which is beneficial to providing a basis for subsequent model update and strategy adjustment.

[0143] The model update condition refers to the condition for triggering the update of the power system dispatching model, so that the server can further update the model or re-select a suitable model to generate the dispatching strategy. In specific implementation, the model update condition can be that the dispatching difference exceeds the set threshold, or that the power system does not reach the expected state after running the dispatching strategy. When the dispatching difference meets the model update condition, it indicates that the current dispatching strategy or model cannot accurately reflect the actual operating state of the power system and needs to be updated and adjusted.

[0144] The operation data after dispatching execution refers to various operation data collected by the power system after dispatching execution. When the model update condition is met, the operation data of the power system at this time can be collected as real-time operation data, and the steps of returning to obtain the real-time operation data of the power system and determining the operating state of the power system based on the real-time operation data can be performed to re-execute the entire solution and re-determine the preselected dispatching model; or the steps of returning to determine the matching degree between the operating state and each of the at least one preselected dispatching model based on the real-time operation data and the at least one preselected dispatching model can be performed to re-determine the matching degree between the preselected dispatching model and the operating state and re-determine the target dispatching model.

[0145] In this embodiment, by obtaining the scheduling execution result and determining the scheduling difference from the expected scheduling result, it is possible to determine whether the scheduling strategy conforms to the current operating state of the power system. When the scheduling difference meets the model update condition, the real-time operating data is determined using the operating data after scheduling, so as to re-determine the preselected scheduling model or re-determine the matching degree between the preselected scheduling model and the operating state, ensuring that a more suitable scheduling model can be selected or the parameters and structure of the existing model can be adjusted to improve the accuracy and effectiveness of the scheduling strategy.

[0146] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless clearly stated in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least a part of other steps or steps or stages in other steps.

[0147] Based on the same inventive concept, an embodiment of the present application also provides an uncertainty-based power system scheduling device for implementing the above-described uncertainty-based power system scheduling method. The implementation solutions provided by this device to solve problems are similar to the implementation solutions described in the above method. Therefore, the specific limitations in one or more embodiments of the uncertainty-based power system scheduling device provided below can refer to the limitations on the uncertainty-based power system scheduling method in the above text, and will not be repeated here.

[0148] In an exemplary embodiment, as Figure 4 shown, an uncertainty-based power system scheduling device is provided, including: a state determination module 402, a model screening model 404, a matching degree determination module 406, a model matching module 408, and an operation scheduling module 410, where:

[0149] The state determination module 402 is configured to obtain the real-time operating data of the power system and determine the operating state of the power system according to the real-time operating data;

[0150] The model screening model 404 is configured to determine at least one preselected scheduling model from at least one pre-trained candidate scheduling model according to the state type of the operating state; the candidate scheduling model is trained based on the uncertainty parameters and historical operating data of the power system, and the uncertainty parameters are used to characterize the fluctuation conditions of the power system in the operating states corresponding to their respective state types;

[0151] A matching degree determination module 406, configured to determine the matching degree between the operating state and each of at least one preselected scheduling model according to real-time operating data and at least one preselected scheduling model;

[0152] A model matching module 408, configured to determine a target scheduling model from at least one preselected scheduling model according to the matching degree between the operating state and each of at least one preselected scheduling model and real-time operating data;

[0153] An operation scheduling module 410, configured to generate a scheduling strategy for the power system through the target scheduling model based on real-time operating data, and schedule the power system according to the scheduling strategy.

[0154] In an optional embodiment, the matching degree determination module 406 is further configured to, for each preselected scheduling model, determine a model performance parameter of the preselected scheduling model according to real-time operating data; the model performance parameter is used to characterize the matching degree between the preselected scheduling model and the corresponding state type; determine a weight coefficient of the model performance parameter according to the relative importance of the model performance parameter when the scheduling strategy is generated for the preselected scheduling model; and determine the matching degree between the preselected scheduling model and the operating state according to the model performance parameter and the weight coefficient corresponding to the model performance parameter.

[0155] In an optional embodiment, the model performance parameter includes at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model. The matching degree determination module 406 is further configured to determine at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model according to real-time operating data.

[0156] In an optional embodiment, the matching degree determination module 406 is further configured to respectively determine the time complexity of the preselected scheduling model in terms of running time and the space complexity in terms of storage space according to the data scale of the real-time operating data and the model structure of the preselected scheduling model, and determine the computational complexity according to the time complexity and the space complexity; or determine the running prediction data of the preselected scheduling model at the running moment of the real-time operating data, and determine the model accuracy according to the difference between the real-time operating data and the running prediction data; or determine the preselected strategies of the power system under different fluctuation conditions through the preselected scheduling model based on the real-time operating data, and determine the model stability parameter according to the fluctuation range of each preselected strategy; or determine the adaptability index of the preselected scheduling model in the operating state based on the real-time operating data, and determine the model adaptability parameter according to the adaptability index and the index weight corresponding to the adaptability index.

[0157] In an alternative embodiment, the model matching module 408 is further configured to determine an adaptation parameter between the preselected scheduling model and the operating state according to the matching degree between the operating state and the preselected scheduling model and the real-time operating data; when the adaptation parameter meets the model parameter update condition, iteratively update the model parameters of the preselected scheduling model until the end condition is met to obtain the global optimization parameter; and determine the target scheduling model from at least one preselected scheduling model according to the global optimization parameter corresponding to each of the at least one preselected scheduling models.

[0158] In an alternative embodiment, the power system scheduling device based on uncertainty further includes a model update module, configured to obtain a scheduling execution result of scheduling the power system based on a scheduling policy, and determine a scheduling difference between the scheduling execution result and the expected scheduling result; when the scheduling difference meets the model update condition, determine the operating data of the power system after the scheduling execution; determine the real-time operating data of the power system according to the operating data after the scheduling execution, and return to the step of obtaining the real-time operating data of the power system and determining the operating state of the power system according to the real-time operating data, or return to the step of determining the matching degree between the operating state and each of at least one preselected scheduling models according to the real-time operating data and at least one preselected scheduling model.

[0159] Each module in the above power system scheduling device based on uncertainty can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so as to facilitate the processor to call and execute the operations corresponding to each of the above modules.

[0160] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as Figure 5As shown in the figure. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as the operation data, operation status, scheduling model, and weight coefficient of the power system. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it implements a power system scheduling method based on uncertainty.

[0161] Those skilled in the art can understand that Figure 5 the structure shown in the figure is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0162] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, it implements the power system scheduling method based on uncertainty in each of the above embodiments.

[0163] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, it implements the power system scheduling method based on uncertainty in each of the above embodiments.

[0164] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by the processor, it implements the power system scheduling method based on uncertainty in each of the above embodiments.

[0165] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0166] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0167] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope recorded in this application.

[0168] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A power system scheduling method based on uncertainty, characterized in that The method includes: Obtaining real-time operation data of the power system, and determining the operation state of the power system according to the real-time operation data; Determining at least one preselected scheduling model from at least one pre-trained candidate scheduling model according to the state type of the operation state; the candidate scheduling model is trained based on the uncertainty parameters and historical operation data of the power system, and the uncertainty parameters are used to characterize the fluctuation of the power system in the operation state corresponding to each state type; Determining the matching degree between the operation state and each of the at least one preselected scheduling model according to the real-time operation data and the at least one preselected scheduling model; Determining a target scheduling model from the at least one preselected scheduling model according to the matching degree between the operation state and each of the at least one preselected scheduling model and the real-time operation data; Generating a scheduling strategy for the power system through the target scheduling model based on the real-time operation data, and scheduling the power system according to the scheduling strategy.

2. The method according to claim 1, wherein The determining the matching degree between the operation state and each of the at least one preselected scheduling model according to the real-time operation data and the at least one preselected scheduling model includes: For each of the preselected scheduling models, determining the model performance parameters of the preselected scheduling model according to the real-time operation data; the model performance parameters are used to characterize the matching degree between the preselected scheduling model and the corresponding state type; Determining the weight coefficient of the model performance parameters according to the relative importance of the model performance parameters when the scheduling strategy is generated by the preselected scheduling model; Determining the matching degree between the preselected scheduling model and the operation state according to the model performance parameters and the weight coefficient corresponding to the model performance parameters.

3. The method according to claim 2, wherein The model performance parameters include at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model; The determining the model performance parameters of the preselected scheduling model according to the real-time operation data includes: Determining at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model according to the real-time operation data.

4. The method according to claim 3, wherein The determining at least one of the computational complexity, model accuracy, model stability parameter, and model adaptability parameter of the preselected scheduling model according to the real-time operation data includes at least one of the following: According to the data scale of the real-time operation data and the model structure of the preselected scheduling model, respectively determining the time complexity of the preselected scheduling model in terms of running time and the space complexity in terms of storage space, and determining the computational complexity according to the time complexity and the space complexity; Determining the running prediction data of the preselected scheduling model at the running moment of the real-time operation data, and determining the model accuracy according to the difference between the real-time operation data and the running prediction data; Based on the preselected scheduling model, determine the preselected strategies of the power system under different fluctuation conditions according to the real-time operation data, and determine the model stability parameters according to the fluctuation ranges of the respective preselected strategies; Based on the real-time operation data, determine the adaptability index of the preselected scheduling model in the operation state, and determine the model adaptability parameters according to the adaptability index and the index weights corresponding to the respective adaptability indexes.

5. The method according to claim 2, characterized in that, The determining the target scheduling model from the at least one preselected scheduling model according to the matching degree of the operation state and the respective preselected scheduling models and the real-time operation data includes: Determine the adaptation parameters of the preselected scheduling model and the operation state according to the matching degree of the operation state and the preselected scheduling model and the real-time operation data; When the adaptation parameters meet the model parameter update condition, iteratively update the model parameters of the preselected scheduling model until the end condition is met to obtain the global optimization parameters; Determine the target scheduling model from the at least one preselected scheduling model according to the global optimization parameters corresponding to the respective at least one preselected scheduling models.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Obtain the scheduling execution result of scheduling the power system based on the scheduling strategy, and determine the scheduling difference between the scheduling execution result and the expected scheduling result; When the scheduling difference meets the model update condition, determine the operation data of the power system after scheduling execution; Determine the real-time operation data of the power system according to the operation data after scheduling execution, and return to the step of obtaining the real-time operation data of the power system and determining the operation state of the power system according to the real-time operation data, or return to the step of determining the matching degree of the operation state and the respective at least one preselected scheduling models according to the real-time operation data and the at least one preselected scheduling model.

7. A power system scheduling device based on uncertainty, characterized in that, The device includes: A state determination module, configured to obtain the real-time operation data of the power system and determine the operation state of the power system according to the real-time operation data; A model screening model, configured to determine at least one preselected scheduling model from at least one pre-trained candidate scheduling model according to the state type of the operation state; the candidate scheduling model is trained based on the uncertainty parameters and historical operation data of the power system, and the uncertainty parameters are used to characterize the fluctuation conditions of the power system in the operation states corresponding to their respective state types; A matching degree determination module, configured to determine the matching degree of the operation state and the respective at least one preselected scheduling models according to the real-time operation data and the at least one preselected scheduling model; A model matching module, configured to determine the target scheduling model from the at least one preselected scheduling model according to the matching degree of the operation state and the respective at least one preselected scheduling models and the real-time operation data; The operation scheduling module is used to generate a scheduling strategy for the power system through the target scheduling model based on the real-time operation data, and schedule the power system according to the scheduling strategy.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 6.

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

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method described in any one of claims 1 to 6.

Images