Pipeline-based spot electricity price prediction method and device and storage medium

Abstract

The invention relates to a pipeline-based spot electricity price prediction method and device and a storage medium, and is applied to the technical field of electricity market spot electricity price prediction, and the method comprises the steps: initializing populations on the premise of meeting the updating condition of an original modeling pipeline, screening the initialized populations, and obtaining a primary parent; sequentially mutating the primary parent according to a preset mutation method to obtain a new population; the process is optimized through an automatic genetic algorithm, so that the requirement of manual intervention is reduced, and the construction and maintenance efficiency of modeling pipeline is improved; a wider solution space can be explored by utilizing the global search capability of the genetic algorithm, so that a better prediction model configuration can be found possibly; the method can automatically adapt to market change and data fluctuation by judging whether an updating condition is met or not through the power market boundary data, updates the model pipe in real time, and ensures that the prediction model is always kept synchronous with the latest market condition.

Description

Technical Field

[0001] The present invention relates to the technical field of spot electricity price prediction in the power market, and in particular to a pipeline-based spot electricity price prediction method, device and storage medium. Background Art

[0002] As a key component of power market operations, spot electricity price forecasting technology is vital for power companies, traders, and policymakers. Accurate electricity price forecasts can help market participants make more reasonable power generation plans, scheduling strategies, and trading decisions, thereby reducing costs, optimizing resource allocation, and improving market efficiency; in the electricity spot market, electricity prices are affected by many factors, including supply and demand, weather changes, seasonal factors, policy adjustments, etc. The impact of these factors on electricity prices is complex and unstable, so traditional forecasting methods, such as time series analysis, statistical models, and machine learning algorithms, often require a lot of manual intervention and regular maintenance to adapt to changes in the impact of market boundaries on electricity prices. However, when faced with frequently changing market conditions, the cost of manual detection and maintenance may be too high, and it may not be updated in a timely manner, resulting in a decrease in forecast accuracy. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a pipeline-based spot electricity price forecasting method, device and storage medium to solve the problem in the prior art that traditional forecasting methods often require a large amount of manual intervention and regular maintenance to adapt to changes in the impact of market boundaries on electricity prices, the cost of manual inspection and maintenance may be too high, and may not be updated in time, resulting in a decrease in forecast accuracy.

[0004] According to a first aspect of an embodiment of the present invention, a pipeline-based spot electricity price prediction method is provided, the method comprising:

[0005] Obtain the power market boundary data and determine whether it meets the update conditions of the original modeling pipeline;

[0006] If the update conditions of the original modeling pipeline are met, the population is initialized, and each individual in the initialized population is a complete and executable modeling pipeline;

[0007] Screening the initialized population to obtain the first generation of parents;

[0008] The first generation parents are mutated in sequence according to the preset mutation method to obtain a new population;

[0009] Performing a spot electricity price forecasting task on each individual of the new population on the training set, and selecting a first preset number of optimal individuals according to a preset forecasting evaluation index;

[0010] The first preset number of optimal individuals screened out are continuously mutated in sequence according to the preset mutation method to obtain a new population, and the iteration is repeated until the preset number of iterations is met or the preset test evaluation index is met and the optimization is stopped;

[0011] Obtain the best individual in the training set from the new population obtained in the final iteration;

[0012] The best performing individual is verified, and if the verification passes, the best performing individual is used to update the original modeling pipeline.

[0013] Preferably,

[0014] The step of obtaining the power market boundary data and determining whether the update conditions of the original modeling pipeline are met includes:

[0015] Obtaining power market boundary data, and determining whether the power market boundary data deviates from the spot electricity price distribution;

[0016] If the power market boundary data does not deviate from the spot electricity price distribution, the update condition of the original modeling pipeline is not met;

[0017] If the electricity market boundary data deviates from the spot electricity price distribution, the update condition of the original modeling pipeline is met.

[0018] Preferably,

[0019] The screening of the initialization population to obtain the first generation of parents includes:

[0020] Each individual in the initialized population performs a spot electricity price prediction task on the training set, obtains a prediction result for each individual, compares the prediction result with a preset prediction evaluation index, and selects a second preset number of individuals as initial generation parents.

[0021] Preferably,

[0022] The obtained first-generation parents are mutated in sequence according to a preset mutation method to obtain a new population, including:

[0023] Define a primitive set, which contains all possible operation nodes in the modeling pipeline;

[0024] Performing a proliferation mutation operation on each individual of the primary parent through the Primitive Set;

[0025] Performing an interference mutation operation on each individual of the first-generation parent;

[0026] Performing a primitive mutation operation on each individual of the first-generation parent through the Primitive Set;

[0027] Each individual of the first generation parent body traverses the proliferation mutation operation, the interference mutation operation and the native mutation operation in a random order to obtain the new population.

[0028] Preferably,

[0029] The performing of the proliferation mutation operation on each individual of the primary parent through the Primitive Set includes:

[0030] Select any individual from the first-generation parent, obtain any operation node from the Primitive Set, splice the selected arbitrary operation node before or after the random node of the arbitrary individual, and complete the proliferation mutation operation of the arbitrary individual; repeat this operation until the proliferation mutation operation of each individual of the first-generation parent is completed.

[0031] Preferably,

[0032] The performing of the interference mutation operation on each individual of the first generation parent includes:

[0033] Randomly match the infection pairs of the first-generation parents to obtain n pairs of infection pairs, randomly exchange any nodes between two individuals in each pair of infection pairs, and complete the interference mutation operation of one team of infection pairs; repeat this operation until the interference mutation operation of n pairs of infection pairs is completed;

[0034] If the number of individuals in the first generation parent is an odd number, the remaining individual does not perform the interference mutation operation.

[0035] Preferably,

[0036] The primitive mutation operation is performed on each individual of the first generation parent through the primitive set to obtain a new population including:

[0037] Select any individual from the first-generation parent, obtain any operation node from the Primitive Set, replace the random node of the arbitrary individual by the selected arbitrary operation node, and complete the native mutation operation of the arbitrary individual; repeat this operation until the native mutation operation of each individual of the first-generation parent is completed.

[0038] Preferably, it also includes:

[0039] If the input and output of the mutation node after any individual's proliferation mutation operation, infection mutation operation or native mutation operation does not match the input and output of the previous and next nodes, the proliferation mutation operation, infection mutation operation or native mutation operation is abandoned.

[0040] According to a second aspect of an embodiment of the present invention, a pipeline-based spot electricity price prediction device is provided, the device comprising:

[0041] Update judgment module: used to obtain power market boundary data and determine whether the update conditions of the original modeling pipeline are met;

[0042] Initialization module: used to initialize the population if the update conditions of the original modeling pipeline are met. Each individual in the initialized population is a complete and executable modeling pipeline.

[0043] The first generation parent acquisition module is used to screen the initialization population and obtain the first generation parent;

[0044] Mutation module: used to mutate the first-generation parents in sequence according to the preset mutation method to obtain a new population;

[0045] Screening module: used for executing the spot electricity price forecasting task on each individual of the new population on the training set, and screening a first preset number of optimal individuals according to preset forecasting evaluation indicators;

[0046] Iteration module: used to continue to mutate the first preset number of optimal individuals screened out in sequence according to the preset mutation method to obtain a new population, and repeat the iteration until the preset number of iterations is met or the preset test evaluation index is met and the optimization is stopped;

[0047] Best individual acquisition module: used to obtain the best individual in the training set from the new population obtained in the final iteration;

[0048] Update module: used to verify the best performing individual, and if the verification passes, use the best performing individual to update the original modeling pipeline.

[0049] According to a third aspect of an embodiment of the present invention, there is provided a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a main controller, each step in the above method is implemented.

[0050] The technical solution provided by the embodiments of the present invention may have the following beneficial effects:

[0051] The present application initializes the population and screens the initialized population to obtain the first-generation parents under the premise of satisfying the update conditions of the original modeling pipeline; mutates the first-generation parents in turn according to a preset mutation method to obtain a new population; reduces the need for manual intervention and improves the construction and maintenance efficiency of the modeling pipeline through an automated genetic algorithm optimization process; utilizes the global search capability of the genetic algorithm to explore a wider solution space, thereby making it possible to find a better prediction model configuration; determines whether the update conditions are met through the power market boundary data, can automatically adapt to market changes and data fluctuations, and updates the model pipeline in real time to ensure that the prediction model is always synchronized with the latest market conditions.

[0052] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0054] Figure 1 is a flow chart of a pipeline-based spot electricity price prediction method according to an exemplary embodiment;

[0055] Figure 2 is a schematic diagram of the overall framework of a pipeline-based spot electricity price prediction method according to another exemplary embodiment;

[0056] Figure 3 is a system schematic diagram of a pipeline-based spot electricity price prediction device according to another exemplary embodiment;

[0057] In the attached figure: 1-update judgment module, 2-initialization module, 3-first generation parent acquisition module, 4-mutation module, 5-screening module, 6-iteration module, 7-optimal individual acquisition module, 8-update module. DETAILED DESCRIPTION

[0058] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0059] Embodiment 1

[0060] Figure 1 is a flow chart of a pipeline-based spot electricity price prediction method according to an exemplary embodiment. Figure 1 As shown, the method includes:

[0061] S1, obtain the power market boundary data and determine whether it meets the update conditions of the original modeling pipeline;

[0062] S2, if the update conditions of the original modeling pipeline are met, the population is initialized, and each individual in the initialized population is a complete and executable modeling pipeline;

[0063] S3, screening the initialized population to obtain the first generation of parents;

[0064] S4, mutating the obtained first-generation parents in sequence according to a preset mutation method to obtain a new population;

[0065] S5, performing a spot electricity price forecasting task on each individual of the new population on the training set, and selecting a first preset number of optimal individuals according to a preset forecasting evaluation index;

[0066] S6, continuing to mutate the first preset number of optimal individuals screened out in sequence according to a preset mutation method to obtain a new population, and repeating the iteration until a preset number of iterations is met or a preset test evaluation index is met and the optimization is stopped;

[0067] S7, obtain the individual with the best performance in the training set in the new population obtained by the final iteration;

[0068] S8, verifying the best performing individual, and if the verification passes, using the best performing individual to update the original modeling pipeline;

[0069] It is understandable that if Figure 2As shown, the present application obtains market boundary data, including: supply and demand forecast: supply and demand forecast is an important part of the boundary data of the power market, which reflects the supply and demand situation of the power market and plays a key role in market clearing and price forecasting; operation mode: the operation mode of the power system directly affects the supply and demand balance and price fluctuations in the market, and different operation modes will have different impacts on the boundary conditions of the power market; constraints: the constraints of the power market include equipment constraints and system constraints, equipment constraints such as the maximum and minimum output constraints of the generator set, the ramp rate constraints, etc.; system constraints such as node voltage constraints, system frequency constraints, etc.; quotation scheme: the quotation scheme of market participants is also an important part of the boundary data, which reflects the competition strategy and cost structure of market participants, and has a direct impact on the market clearing results, etc.; when the power market boundary data is offset from the spot electricity price distribution adopted by the original model pipeline, it indicates that the modeling pipeline needs to be adjusted, that is, the update step of the modeling pipeline is triggered. The modeling pipeline refers to the process of converting raw data into valuable information and models through a series of orderly steps and tools during data processing, analysis and modeling. The update steps of the modeling pipeline include:

[0070] Initialize the population. Each individual is a potential, complete and executable pipeline. That is, each individual can use the training set data to predict spot electricity. It should be noted that each individual contains multiple operation nodes, and the number of operation nodes contained in each individual may be different. In short, to perform spot electricity prediction, the most basic operation nodes may be only three. As long as these three basic operation nodes are included, the prediction function can be realized. However, other operation nodes can improve the accuracy of model prediction. Therefore, the number of operation nodes of each individual is different, which is why the subsequent proliferation mutation operation can be performed; perform the prediction task on each individual on the training set, and select a certain number of optimal individuals as the first generation of parents according to the prediction evaluation index;

[0071] Use mutations to generate new populations, simulate the natural evolution of organisms, conduct a series of mutation explorations, and evolve a population suitable for survival (i.e. pipeline) through the survival of the fittest. In order to fit the debugging ideas of machine learning modeling, mutation operations include three types: proliferation mutation, infection mutation, and native mutation;

[0072] Before performing the above mutation operation, you also need to define the Primitive Set, which needs to include all possible operation nodes in the modeling pipeline. The operation nodes must at least include data processing methods, feature generation methods, feature selection, machine learning model selection, and hyperparameter settings of the corresponding model; an operation node is a specific option in a type of node task. For example, in a machine learning model, selecting the Xgboost model is an operation, and in feature selection, selecting feature set A is an operation;

[0073] Proliferative mutation: randomly select a new operation from the primitive set and splice it before or after a random node of the individual. This mutation increases the depth of the individual tree structure, that is, explores the complexity of the pipeline;

[0074] Infection mutation: Randomly select two individuals in the current population and exchange any of their nodes; this mutation can increase the diversity of the population, that is, broaden the range of exploration pipeline combinations;

[0075] Native mutation: randomly select a new operation from the primitive set and replace it with a random node of the individual. The purpose of this mutation is to explore different operation combinations while maintaining the complexity of the pipeline;

[0076] It is worth emphasizing that if the input and output of the mutated node do not match the input and output of the previous and next nodes, the mutation needs to be abandoned;

[0077] The above three mutation types are traversed in random order, and all individuals in the current population are mutated; for example, when performing a proliferation mutation, each individual in the population will be traversed for a proliferation mutation, and then the next mutation type will be started. For infection mutations, the algorithm first randomly matches infection pairs for the current population, and then traverses the infection pairs to complete the mutation. If the number of individuals in the population is an odd number, the remaining individual will not mutate; after this step is completed, a new population will be generated and enter the next step.

[0078] Each individual in the new population performs the prediction task on the training set, selects a certain number of optimal individuals based on the prediction evaluation index, and continues to perform mutation iterations until a certain number of iterations are completed or the test evaluation index stops optimizing;

[0079] Output the best performing individuals in the training set in the final population. The best performing individuals need to be verified. The verification method can be to compare the prediction results of the original modeling pipeline with the prediction results of the best performing individuals selected after iterative update. If the prediction results of the best performing individuals selected after iterative update are more accurate than the prediction results of the original modeling pipeline, it indicates that the verification has passed. If the verification passes, the original modeling pipeline is updated with the best performing individuals selected after iterative update.

[0080] It is worth noting that, in the solution of the present application, in addition to determining the update conditions to trigger the update of the modeling pipeline, it can also be manually triggered.

[0081] Embodiment 2

[0082] Figure 3 is a system schematic diagram of a pipeline-based spot electricity price prediction device according to another exemplary embodiment, the device comprising:

[0083] Update judgment module 1: used to obtain the power market boundary data and determine whether the update conditions of the original modeling pipeline are met;

[0084] Initialization module 2: used to initialize the population if the update conditions of the original modeling pipeline are met, and each individual in the initialized population is a complete and executable modeling pipeline;

[0085] The first generation parent acquisition module 3 is used to screen the initialization population and obtain the first generation parent;

[0086] Mutation module 4: used to mutate the first-generation parents in sequence according to the preset mutation method to obtain a new population;

[0087] Screening module 5: used for executing the spot electricity price forecasting task on each individual of the new population on the training set, and screening a first preset number of optimal individuals according to a preset forecasting evaluation index;

[0088] Iteration module 6: used to continue to mutate the first preset number of optimal individuals screened out in sequence according to the preset mutation method to obtain a new population, and repeat the iteration until the preset number of iterations is met or the preset test evaluation index is met and the optimization is stopped;

[0089] Best individual acquisition module 7: used to obtain the best individual in the training set from the new population obtained in the final iteration;

[0090] Update module 8: used to verify the best performing individual, and if the verification passes, use the best performing individual to update the original modeling pipeline.

[0091] Embodiment three:

[0092] This embodiment provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a main controller, each step in the above method is implemented;

[0093] It is understandable that the storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0094] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0095] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0096] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0097] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0098] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0099] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0100] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0101] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0102] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. The pipeline-based spot electricity price prediction method is characterized by: The method comprises: Obtain the power market boundary data and determine whether it meets the update conditions of the original modeling pipeline; If the update conditions of the original modeling pipeline are met, the population is initialized, and each individual in the initialized population is a complete and executable modeling pipeline; Screening the initialized population to obtain the first generation of parents; The first generation parents are mutated in sequence according to the preset mutation method to obtain a new population; Performing a spot electricity price forecasting task on each individual of the new population on the training set, and selecting a first preset number of optimal individuals according to a preset forecasting evaluation index; The first preset number of optimal individuals screened out are continuously mutated in sequence according to a preset mutation method to obtain a new population, and the iteration is repeated until a preset number of iterations is met or a preset test evaluation index is met and the optimization is stopped; Obtain the best individual in the training set from the new population obtained in the final iteration; The best performing individual is verified, and if the verification passes, the best performing individual is used to update the original modeling pipeline.

2. The method according to claim 1, characterized in that The step of obtaining the power market boundary data and determining whether the update conditions of the original modeling pipeline are met includes: Obtaining power market boundary data, and determining whether the power market boundary data deviates from the spot electricity price distribution; If the power market boundary data does not deviate from the spot electricity price distribution, the update condition of the original modeling pipeline is not met; If the electricity market boundary data deviates from the spot electricity price distribution, the update condition of the original modeling pipeline is met.

3. The method according to claim 2, characterized in that The screening of the initialization population to obtain the first generation of parents includes: Each individual in the initialized population performs a spot electricity price prediction task on the training set, obtains a prediction result for each individual, compares the prediction result with a preset prediction evaluation index, and selects a second preset number of individuals as initial generation parents.

4. The method according to claim 3, characterized in that The obtained first-generation parents are mutated in sequence according to a preset mutation method to obtain a new population, including: Define a primitive set, which contains all possible operation nodes in the modeling pipeline; Performing a proliferation mutation operation on each individual of the primary parent through the Primitive Set; Performing an interference mutation operation on each individual of the first-generation parent; Performing a primitive mutation operation on each individual of the first-generation parent through the Primitive Set; Each individual of the first generation parent body traverses the proliferation mutation operation, the interference mutation operation and the native mutation operation in a random order to obtain the new population.

5. The method according to claim 4, characterized in that The performing of the proliferation mutation operation on each individual of the primary parent through the Primitive Set includes: Select any individual from the first-generation parent, obtain any operation node from the Primitive Set, splice the selected arbitrary operation node before or after the random node of the arbitrary individual, and complete the proliferation mutation operation of the arbitrary individual; repeat this operation until the proliferation mutation operation of each individual of the first-generation parent is completed.

6. The method according to claim 5, characterized in that The performing of the interference mutation operation on each individual of the first generation parent includes: Randomly match the infection pairs of the first-generation parents to obtain n pairs of infection pairs, randomly exchange any nodes between two individuals in each pair of infection pairs, and complete the interference mutation operation of one team of infection pairs; repeat this operation until the interference mutation operation of n pairs of infection pairs is completed; If the number of individuals in the first generation parent is an odd number, the remaining individual does not perform the interference mutation operation.

7. The method according to claim 6, characterized in that The primitive mutation operation is performed on each individual of the first generation parent through the primitive set to obtain a new population including: Select any individual from the first-generation parent, obtain any operation node from the Primitive Set, replace the random node of the arbitrary individual by the selected arbitrary operation node, and complete the native mutation operation of the arbitrary individual; repeat this operation until the native mutation operation of each individual of the first-generation parent is completed.

8. The method according to any one of claims 5 to 7, characterized in that: Also includes: If the input and output of the mutation node after any individual's proliferation mutation operation, infection mutation operation or native mutation operation does not match the input and output of the previous and next nodes, the proliferation mutation operation, infection mutation operation or native mutation operation is abandoned.

9. The spot electricity price prediction device based on pipeline is characterized by: The device comprises: Update judgment module: used to obtain power market boundary data and determine whether the update conditions of the original modeling pipeline are met; Initialization module: used to initialize the population if the update conditions of the original modeling pipeline are met. Each individual in the initialized population is a complete and executable modeling pipeline. The first generation parent acquisition module is used to screen the initialization population and obtain the first generation parent; Mutation module: used to mutate the first-generation parents in sequence according to the preset mutation method to obtain a new population; Screening module: used for executing the spot electricity price forecasting task on each individual of the new population on the training set, and screening a first preset number of optimal individuals according to preset forecasting evaluation indicators; Iteration module: used to continue to mutate the first preset number of optimal individuals screened out in sequence according to the preset mutation method to obtain a new population, and repeat the iteration until the preset number of iterations is met or the preset test evaluation index is met and the optimization is stopped; Best individual acquisition module: used to obtain the best individual in the training set from the new population obtained in the final iteration; Update module: used to verify the best performing individual, and if the verification passes, use the best performing individual to update the original modeling pipeline.

10. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the main controller, each step in the pipeline-based spot electricity price prediction method according to any one of claims 1 to 8 is implemented.

Images