Electronic device for predicting amount of power generation on basis of dimensional transformation for multi-cycle time series data and operating method thereof

Abstract

The present disclosure relates to an operating method of an electronic device. The operating method identifies a power generation cycle on the basis of the amount of power generation over time, and predicts the amount of power generation on the basis of a pattern in the cycle and a pattern between cycles for the power generation cycle on the basis of dimensional transformation for multi-cycle time series data. Specifically, the method comprises the steps of: acquiring 1-dimension (1D) power generation data including the amount of power generation over time; acquiring, on the basis of the 1D power generation data, 2-dimension (2D) matrix data including at least one power generation cycle and the amplitude related to the amount of power generation corresponding to the power generation cycle; and predicting the amount of power generation for at least one target time point on the basis of feature information about the 2D matrix data.

Description

Electronic device for predicting power generation based on dimension transformation of multi-period time series data and method of operation thereof

[0001] The present disclosure relates to an electronic device and a method of operating the same, and more specifically, to an electronic device and a method of operating the same that identify a power generation cycle based on a power generation amount over time and predict a power generation amount based on patterns within the cycle and patterns between cycles regarding the power generation cycle.

[0002] As issues of climate change and energy supply instability intensify recently, the importance of carbon neutrality is being highlighted globally, and the utilization of new and renewable energy to achieve this is emerging as an essential task.

[0003] However, power generation using renewable energy has the problem of significant variability in output depending on various conditions, so if this cannot be effectively predicted, difficulties arise in power management and planning.

[0004] Conventional technologies that use prediction models such as LSTM (Long Short-Term Memory), GRU (Gated Recurrent Unit), and Transformer to predict power generation have limitations in learning complex patterns from time series data, and prediction performance often degrades due to overfitting.

[0005] Therefore, there is a need to propose a prediction system that improves the accuracy and stability of power generation forecasting by appropriately reflecting the uncertainties of various variables.

[0006] The present disclosure aims to provide an electronic device and a method of operation thereof that identify a power generation cycle based on the amount of power generated over time, and predict the amount of power generated based on patterns within the cycle and patterns between cycles regarding the power generation cycle.

[0007] The purposes of the present disclosure are not limited to those mentioned above, and other purposes and advantages of the present disclosure not mentioned may be understood from the following description and will be more clearly understood from the embodiments of the present disclosure. Furthermore, it will be readily apparent that the purposes and advantages of the present disclosure can be realized by the means and combinations thereof set forth in the claims.

[0008] A method of operating an electronic device according to one embodiment of the present disclosure includes the steps of: acquiring 1D power generation data including a power generation amount over time; acquiring 2D matrix data including at least one power generation cycle and an amplitude related to a power generation amount corresponding to the power generation cycle based on the 1D power generation data; and predicting a power generation amount for at least one target time point based on feature information of the 2D matrix data.

[0009] Here, the step of acquiring the 2D matrix data may include the step of acquiring transformed data including frequency and amplitude related to the amount of power generated by performing a Fast Fourier Transform on the 1D power generation data, and the step of generating the 2D matrix data based on the transformed data.

[0010] At this time, the step of generating the 2D matrix data may be to identify a frequency corresponding to each of a plurality of amplitudes having an amplitude value exceeding a reference value from the transformed data, identify a plurality of generation cycles based on the identified frequency, arrange a pattern within the cycle indicating a phase difference between different points in time within the same generation cycle among the identified plurality of generation cycles in rows, and arrange a pattern between the periods indicating a time difference between points having the same phase within different generation cycles in columns to generate the 2D matrix data.

[0011] At this time, the step of predicting the amount of power generated for at least one target time point may include the step of applying at least one convolution layer to the 2D matrix data to extract feature information related to the pattern within the period and the pattern between the periods, the step of converting the extracted feature information into a one-dimensional form, and the step of predicting the amount of power generated for the target time point based on the feature information converted into a one-dimensional form.

[0012] At this time, the step of converting the extracted feature information into a one-dimensional form can be performed by connecting patterns corresponding to each of at least one row and at least one column constituting the extracted feature information to convert it into a one-dimensional form.

[0013] At this time, the step of predicting the amount of power generated for the target time point based on the feature information converted into the one-dimensional form can be performed by identifying changes within the cycle and changes between periods for the plurality of power generation cycles based on the feature information converted into the one-dimensional form and the 2D matrix data, and obtaining the amount of power generated predicted for the target time point according to the power generation cycle corresponding to the target time point based on the identified changes within the cycle and changes between periods.

[0014] Meanwhile, the method of operating the electronic device may include the steps of: obtaining a predicted power generation history including a predicted power generation amount for each of a plurality of time points and a power generation history including an actual power generation amount actually generated for each of the plurality of time points; identifying a time point where the difference value between the predicted power generation amount and the actual power generation amount exceeds a threshold value based on the predicted power generation history and the power generation history; calculating the time between the time point where the power generation amount for the identified time point is predicted and the time point where the identified time point is set as a threshold time; and setting at least one of the time points within the threshold time from the time point where the power generation amount is predicted as the target time point.

[0015] An electronic device according to one embodiment of the present disclosure includes a memory in which a power generation prediction model for predicting power generation over time is stored, and a processor connected to the memory and which inputs 1D power generation data including power generation over time into the power generation prediction model and obtains power generation for at least one target time point based on the output data.

[0016] Here, the processor can obtain transformed data including frequency and amplitude related to power generation by performing a Fast Fourier Transform on the 1D power generation data through the power generation prediction model, and through the power generation prediction model, identify a frequency corresponding to each of a plurality of amplitudes having an amplitude value exceeding a reference value from the transformed data, identify a plurality of power generation cycles based on the identified frequency, and obtain 2D matrix data by aligning a pattern within the cycle that indicates a phase difference between different points in time within the same power generation cycle in rows, and aligning a pattern between the cycles that indicates a time difference between points with the same phase within different power generation cycles in columns.

[0017] At this time, the processor can apply at least one convolution layer to the 2D matrix data through the power generation prediction model to extract feature information related to a two-dimensional change over time, convert the feature information into a one-dimensional form by connecting patterns corresponding to each of at least one row and at least one column constituting the extracted feature information, and obtain the power generation amount for the target time point predicted based on the feature information converted into the one-dimensional form and the 2D matrix data through the power generation prediction model.

[0018] A non-transient computer-readable medium according to one embodiment of the present disclosure stores at least one instruction that is executed by a processor of an electronic device to cause the electronic device to perform the method of operation of claim 1.

[0019] Through the present disclosure, stable prediction of power generation is possible, thereby enhancing the stability of power supply, supporting the efficient operation of energy storage devices, and contributing to the reduction of energy costs and carbon emissions.

[0020] In addition, the present disclosure can help with efficient long-term energy management by optimizing the operation plan of an energy storage system based on the predicted power generation amount.

[0021] FIG. 1 is a drawing for explaining the operation of an electronic device predicting power generation amount according to one embodiment of the present disclosure.

[0022] FIG. 2 is a diagram illustrating the operation of an electronic device according to one embodiment of the present disclosure to predict the amount of power generated at a target time based on feature information of 2D matrix data.

[0023] FIG. 3 is a diagram illustrating the process of an electronic device according to one embodiment of the present disclosure extracting features from a tensor reconstructed in two dimensions.

[0024] FIG. 4 is a block diagram illustrating the configuration of an electronic device according to one embodiment of the present disclosure.

[0025] FIG. 5 is a diagram exemplarily illustrating the definition of a power generation prediction model according to one embodiment of the present disclosure.

[0026] FIG. 6 is a diagram illustrating the operation of evaluating the predicted value of a power generation prediction model according to one embodiment of the present disclosure.

[0027] FIG. 7 is a diagram illustrating the operation of an electronic device communicating with a manager device according to one embodiment of the present disclosure.

[0028] A method of operating an electronic device according to one embodiment of the present disclosure includes the steps of: acquiring 1D power generation data including a power generation amount over time; acquiring 2D matrix data including at least one power generation cycle and an amplitude related to a power generation amount corresponding to the power generation cycle based on the 1D power generation data; and predicting a power generation amount for at least one target time point based on feature information of the 2D matrix data.

[0029] Here, the step of acquiring the 2D matrix data may include the step of acquiring transformed data including frequency and amplitude related to the amount of power generated by performing a Fast Fourier Transform on the 1D power generation data, and the step of generating the 2D matrix data based on the transformed data.

[0030] At this time, the step of generating the 2D matrix data may be to identify a frequency corresponding to each of a plurality of amplitudes having an amplitude value exceeding a reference value from the transformed data, identify a plurality of generation cycles based on the identified frequency, arrange a pattern within the cycle indicating a phase difference between different points in time within the same generation cycle among the identified plurality of generation cycles in rows, and arrange a pattern between the periods indicating a time difference between points having the same phase within different generation cycles in columns to generate the 2D matrix data.

[0031] At this time, the step of predicting the amount of power generated for at least one target time point may include the step of applying at least one convolution layer to the 2D matrix data to extract feature information related to the pattern within the period and the pattern between the periods, the step of converting the extracted feature information into a one-dimensional form, and the step of predicting the amount of power generated for the target time point based on the feature information converted into a one-dimensional form.

[0032] At this time, the step of converting the extracted feature information into a one-dimensional form can be performed by connecting patterns corresponding to each of at least one row and at least one column constituting the extracted feature information to convert it into a one-dimensional form.

[0033] At this time, the step of predicting the amount of power generated for the target time point based on the feature information converted into the one-dimensional form can be performed by identifying changes within the cycle and changes between periods for the plurality of power generation cycles based on the feature information converted into the one-dimensional form and the 2D matrix data, and obtaining the amount of power generated predicted for the target time point according to the power generation cycle corresponding to the target time point based on the identified changes within the cycle and changes between periods.

[0034] Meanwhile, the method of operating the electronic device may include the steps of: obtaining a predicted power generation history including a predicted power generation amount for each of a plurality of time points and a power generation history including an actual power generation amount actually generated for each of the plurality of time points; identifying a time point where the difference value between the predicted power generation amount and the actual power generation amount exceeds a threshold value based on the predicted power generation history and the power generation history; calculating the time between the time point where the power generation amount for the identified time point is predicted and the time point where the identified time point is set as a threshold time; and setting at least one of the time points within the threshold time from the time point where the power generation amount is predicted as the target time point.

[0035] An electronic device according to one embodiment of the present disclosure includes a memory in which a power generation prediction model for predicting power generation over time is stored, and a processor connected to the memory and which inputs 1D power generation data including power generation over time into the power generation prediction model and obtains power generation for at least one target time point based on the output data.

[0036] Here, the processor can obtain transformed data including frequency and amplitude related to power generation by performing a Fast Fourier Transform on the 1D power generation data through the power generation prediction model, and through the power generation prediction model, identify a frequency corresponding to each of a plurality of amplitudes having an amplitude value exceeding a reference value from the transformed data, identify a plurality of power generation cycles based on the identified frequency, and obtain 2D matrix data by aligning a pattern within the cycle that indicates a phase difference between different points in time within the same power generation cycle in rows, and aligning a pattern between the cycles that indicates a time difference between points with the same phase within different power generation cycles in columns.

[0037] At this time, the processor can apply at least one convolution layer to the 2D matrix data through the power generation prediction model to extract feature information related to a two-dimensional change over time, convert the feature information into a one-dimensional form by connecting patterns corresponding to each of at least one row and at least one column constituting the extracted feature information, and obtain the power generation amount for the target time point predicted based on the feature information converted into the one-dimensional form and the 2D matrix data through the power generation prediction model.

[0038] A non-transient computer-readable medium according to one embodiment of the present disclosure stores at least one instruction that is executed by a processor of an electronic device to cause the electronic device to perform the method of operation of claim 1.

[0039] Before specifically describing the present disclosure, the method of description in the specification and drawings is described.

[0040] First, the terms used in this specification and claims have been selected based on general terms considering their functions in the various embodiments of this disclosure. However, these terms may vary depending on the intent of those skilled in the art, legal or technical interpretations, and the emergence of new technologies. Additionally, some terms have been arbitrarily selected by the applicant. Such terms may be interpreted according to the meanings defined in this specification; in the absence of specific definitions, they may be interpreted based on the overall content of this specification and common technical knowledge in the relevant field.

[0041] In addition, the same reference numbers or symbols described in each drawing attached to this specification represent parts or components that perform substantially the same function. For convenience of explanation and understanding, the same reference numbers or symbols are used to describe different embodiments. That is, even if components having the same reference number are all depicted in multiple drawings, the multiple drawings do not imply a single embodiment.

[0042] Additionally, in this specification and claims, terms including ordinal numbers, such as "first," "second," etc., may be used to distinguish between components. These ordinal numbers are used to distinguish identical or similar components from one another, and the meaning of the terms should not be limited by the use of such ordinal numbers. For example, the order of use or arrangement of components combined with such ordinal numbers should not be restricted by the number. If necessary, each ordinal number may be used interchangeably.

[0043] In this specification, singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "consisting of" are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0044] In the embodiments of the present disclosure, terms such as "module," "unit," "part," etc. are used to refer to a component that performs at least one function or operation, and such component may be implemented in hardware or software, or in a combination of hardware and software. Additionally, a plurality of "modules," "units," "parts," etc. may be integrated into at least one module or chip and implemented as at least one processor, except where each needs to be implemented in specific individual hardware.

[0045] Furthermore, in the embodiments of the present disclosure, when a part is described as being connected to another part, this includes not only a direct connection but also an indirect connection through another medium. Additionally, the meaning that a part includes a certain component implies that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0046] FIG. 1 is a drawing for explaining the operation of an electronic device predicting power generation amount according to one embodiment of the present disclosure.

[0047] The electronic device (100) can acquire 1D power generation data including the amount of power generated over time (S110).

[0048] In one embodiment, the electronic device (100) can acquire 1D power generation data through user input.

[0049] For example, the electronic device (100) may receive 1D generation data from an external terminal device, or may acquire 1D generation data according to user input through the electronic device (100).

[0050] As an additional embodiment, the electronic device (100) may be connected to a power generation device that generates electricity and obtain power generation amount data generated through the power generation device to obtain 1D power generation data.

[0051] Here, the power generation device may correspond to a device that obtains electrical energy using at least one of various power generation methods, such as a photovoltaic power generation device that obtains solar energy through at least one photovoltaic panel and converts it into electrical energy, a wind power generation device that obtains rotational energy through at least one blade and converts it into electrical energy, a hydroelectric power generation device that produces electrical energy using the potential energy of water, or a geothermal power generation device that produces electrical energy using geothermal energy.

[0052] The electronic device (100) can obtain 2D matrix data including amplitudes related to the power generation cycle and the amount of power generated corresponding to the power generation cycle based on 1D power generation data (S120).

[0053] In one embodiment, the electronic device (100) can obtain transformed data including frequency and amplitude related to the amount of power generated by applying a fast Fourier transform to 1D power generation data.

[0054] Specifically, the electronic device (100) can obtain transformed data having a frequency domain (e.g., data including power generation amount according to frequency) by applying a Fast Fourier Transform (FFT) to 1D power generation data having a time domain (e.g., data including power generation amount according to time).

[0055] Here, the electronic device (100) may perform embedding on the 1D power generation data to perform a Fast Fourier Transform on the 1D power generation data, and then apply the Fast Fourier Transform to the embedded 1D power generation data.

[0056] At this time, the electronic device (100) can identify a frequency corresponding to each of a plurality of amplitudes having an amplitude value exceeding a reference value from the converted data. Here, the amplitude value may be a value corresponding to the amount of power generated.

[0057] For example, the electronic device (100) can set the rank of each amplitude according to the magnitude of the amplitude value included in the conversion data, and set the amplitude value of the amplitude corresponding to the set rank as a reference value.

[0058] Here, the electronic device (100) can identify multiple power generation cycles based on the identified frequency.

[0059] Specifically, the electronic device (100) can set the reciprocal of each identified frequency as the generation cycle.

[0060] At this time, the electronic device (100) can identify a pattern within the cycle that indicates a phase difference between different points in time within the same power generation cycle among a plurality of power generation cycles, and a pattern between the cycles that indicates a time difference between points having the same phase within different power generation cycles.

[0061] Accordingly, the electronic device (100) can obtain 2D matrix data in which the pattern within the period is aligned in rows and the pattern between the period is aligned in columns.

[0062] The electronic device (100) can predict the amount of power generated at a target time based on the feature information of 2D matrix data (S130).

[0063] In one embodiment, the electronic device (100) can extract feature information related to in-period patterns and period-period patterns by applying at least one convolution layer to 2D matrix data.

[0064] For example, an electronic device (100) can extract feature information from 2D matrix data through at least one block comprising a plurality of convolution layers having kernels of a preset size. Here, the plurality of convolution layers may have kernels of different sizes or kernels of the same size, but are not limited thereto.

[0065] For example, the electronic device (100) can identify not only the change cycle based on the difference in power generation between adjacent time points but also the change cycle based on the difference in power generation between power generation cycles by extracting feature information from 2D matrix data in which the pattern within the cycle and the pattern between cycles are arranged in rows and columns.

[0066] At this time, the electronic device (100) can use feature information extracted from 2D matrix data to predict the amount of power generated for a target time.

[0067] A more detailed explanation regarding this will be provided later through Fig. 2.

[0068] As an additional embodiment, the electronic device (100) can generate weather forecast information for a region corresponding to the location information of the power generation device based on the predicted amount of power generation (: predicted amount of power generation) for a plurality of target times.

[0069] Specifically, when the type of power generation device (e.g., solar, wind, hydro, geothermal, thermal, etc.) corresponds to a type related to weather (e.g., solar, wind, hydro, etc.), the electronic device (100) can predict that a weather condition corresponding to the type of power generation device will appear at the time when the amount of power generated by the power generation device is predicted to exceed a preset value.

[0070] Accordingly, the electronic device (100) can generate weather forecast information that matches the time when the predicted amount of power generation exceeds a preset value and the weather conditions corresponding to the type of power generation device.

[0071] For example, if the type of power generation device corresponds to wind power, the electronic device (100) can generate weather forecast information for a region corresponding to the location information of the power generation device by predicting that a strong wind will occur at the time when the amount of power generated by the power generation device exceeds a preset value.

[0072] FIG. 2 is a diagram illustrating the operation of an electronic device according to one embodiment of the present disclosure to predict the amount of power generated at a target time based on feature information of 2D matrix data.

[0073] Referring to FIG. 2, the electronic device (100) can extract feature information related to a two-dimensional change over time from 2D matrix data (S210).

[0074] Specifically, the electronic device (100) can extract feature information related to in-period patterns and period-period patterns by applying at least one convolution layer to 2D matrix data.

[0075] The electronic device (100) can convert the extracted feature information into a one-dimensional form (S220)

[0076] In one embodiment, the electronic device (100) can convert the pattern corresponding to each of at least one row and at least one column constituting the feature information into a one-dimensional form.

[0077] Here, the feature information converted into a one-dimensional form may correspond to a feature map converted into a one-dimensional tensor form by flattening a feature map corresponding to a two-dimensional tensor along the time axis.

[0078] The electronic device (100) can predict the amount of power generated at a target time based on feature information converted into a one-dimensional form (S230).

[0079] In one embodiment, the electronic device (100) can identify changes within a cycle and changes between cycles for a plurality of power generation cycles based on feature information converted into a one-dimensional form and 2D matrix data.

[0080] For example, the electronic device (100) may identify that the first to Nth generation cycles are repeated based on feature information converted into a one-dimensional form and 2D matrix data, and may also identify detailed cycles in which the amount of power generated increases or decreases within the first generation cycle.

[0081] At this time, the electronic device (100) can obtain a predicted amount of power generation for a target time corresponding to a power generation cycle based on changes within a cycle and changes during a cycle for a plurality of power generation cycles.

[0082] For example, the electronic device (100) can predict a power generation cycle corresponding to a target time point based on a change in the period of a plurality of power generation cycles, and predict the amount of power generated for a target time point according to the location where the target time point is located within the power generation cycle corresponding to the target time point based on a change within the period of a plurality of power generation cycles.

[0083] Meanwhile, the electronic device (100) can set a target time based on a predicted power generation history including a predicted amount of power generation and a power generation history including an actual amount of power generated.

[0084] Specifically, the electronic device (100) can obtain a predicted power generation history including a predicted amount of power generated for each of a plurality of time points and a power generation history including an actual amount of power generated for each of the plurality of time points.

[0085] At this time, the electronic device (100) can identify the point in time when the difference between the predicted power generation amount and the actual power generation amount exceeds a threshold value based on the predicted power generation history and the power generation history.

[0086] Here, the electronic device (100) can calculate the time between the time when the amount of power generated for the identified time is predicted and the time when the identified time is identified, and set this time as the threshold time.

[0087] Accordingly, the electronic device (100) may set at least one point in time within a threshold time from the point in time when the amount of power generated is predicted as a target point in time. If multiple threshold times are set, the electronic device (100) may calculate an average value for the multiple threshold times and set at least one point in time within the calculated average value from the point in time when the amount of power generated is predicted as a target point in time.

[0088] For example, the electronic device (100) can set each of the multiple time points included in the time interval between the time point where the amount of power generated is predicted and the time point where the critical time has elapsed as the target time point.

[0089] In addition, the electronic device (100) may set the target time based on user input.

[0090] At this time, if the target time set according to user input corresponds to a time after the threshold time from the time of predicting the amount of power generated, the electronic device (100) provides information on the amount of power generated predicted for the target time (set according to user input) based on the time when the target time is set according to user input, and when a time corresponding to a time before the threshold time from the target time is reached, the amount of power generated for the target time can be predicted again and provided to the user.

[0091] Through this, the electronic device (100) can provide the user with information on the amount of power generated with high prediction accuracy, thereby helping the user optimize the operation plan of the energy storage system and assist in efficient long-term energy management.

[0092] FIG. 3 is a diagram illustrating the process of an electronic device according to one embodiment of the present disclosure extracting features from a tensor reconstructed in two dimensions.

[0093] Referring to FIG. 3, the electronic device (100) can obtain a two-dimensional tensor by converting time series data corresponding to a one-dimensional tensor (1D tensor) into a two-dimensional tensor.

[0094] In one embodiment, the electronic device (100) can convert 1D power generation data including a power generation amount over time to obtain 2D matrix data including at least one power generation cycle and an amplitude associated with a power generation amount corresponding to each power generation cycle.

[0095] Here, the electronic device (100) can apply at least one convolution layer to 2D matrix data. At this time, the convolution layer may include kernels of different sizes or kernels of the same size.

[0096] At this time, the electronic device (100) can convert the acquired feature map into a one-dimensional form by applying a convolution layer to 2D matrix data.

[0097] For example, the electronic device (100) can flatten a feature map corresponding to a two-dimensional tensor and convert it into a one-dimensional tensor.

[0098] Here, when performing shape transformation on a feature map obtained by applying a convolution layer, if padding is applied by the convolution layer applied to the 2D matrix data, the electronic device (100) may remove the padding from the flattened feature map with applied padding and convert it to its original length.

[0099] FIG. 4 is a block diagram illustrating the configuration of an electronic device according to one embodiment of the present disclosure.

[0100] Referring to FIG. 4, the electronic device (100) may include a memory (110) and a processor (120), and the memory (110) may include a power generation prediction model (111).

[0101] The memory (110) is configured to store at least one instruction or data related to an operating system (OS) for controlling the overall operation of the components of the electronic device (100) and the components of the electronic device (100).

[0102] The memory (110) may include non-volatile memory such as ROM or flash memory, and may include volatile memory such as DRAM. Additionally, the memory (110) may include a hard disk, an SSD (Solid state drive), etc.

[0103] In one embodiment, the memory (110) may include a power generation prediction model (111) for predicting the amount of power generated over time.

[0104] For example, the power generation prediction model (111) may correspond to a model trained to convert one-dimensional time series data into two-dimensional image data based on TimesNet, extract temporal and spatial patterns from the one-dimensional time series data and two-dimensional image data, and output a predicted value based on the extracted pattern.

[0105] Specifically, the power generation prediction model (111) performs a fast Fourier transform on 1D power generation data to obtain transformed data including frequency and amplitude related to power generation, and identifies a frequency corresponding to each of a plurality of amplitudes having an amplitude value exceeding a reference value from the transformed data, and can identify a plurality of power generation cycles based on the identified frequency.

[0106] Accordingly, the power generation prediction model (111) can obtain 2D matrix data by aligning the in-cycle pattern, which represents the phase difference between different points in time within the same power generation cycle among multiple power generation cycles, into rows, and aligning the in-cycle pattern, which represents the time difference between points in time having the same phase within different power generation cycles, into columns.

[0107] Here, the power generation prediction model (111) can apply at least one convolution layer to 2D matrix data to extract feature information related to two-dimensional changes over time, and convert the feature information into a one-dimensional form by connecting patterns corresponding to each of at least one row and at least one column constituting the extracted feature information.

[0108] Accordingly, the power generation prediction model (111) can obtain the power generation amount for the predicted target time based on feature information converted into a one-dimensional form and the 2D matrix data.

[0109] The processor (120) is configured to control the electronic device (100) overall.

[0110] In one embodiment, the processor (120) can obtain the amount of power generated for at least one target time point based on the output data obtained by inputting 1D power generation data including the amount of power generated over time into a power generation prediction model (111).

[0111] Meanwhile, the processor (120) may include a general-purpose processor such as a CPU (Central Processing Unit), AP, DSP (Digital Signal Processor), a graphics-dedicated processor such as a GPU (Graphic Processor Unit) or VPU (Vision Processing Unit), or an artificial intelligence-dedicated processor such as an NPU (Neural Processing Unit). An artificial intelligence-dedicated processor may be designed with a hardware structure specialized for training or utilizing a specific artificial intelligence model.

[0112] FIG. 5 is a diagram exemplarily illustrating the definition of a power generation prediction model according to one embodiment of the present disclosure.

[0113] Referring to FIG. 5, the power generation prediction model (111) may include a plurality of convolutional layers and a plurality of fully-connected layers.

[0114] For example, the power generation prediction model (111) can convert the input channel into 32 features through the first convolution layer, expand the 32 features to 64 through the second convolution layer, reduce the 64 features to 32 through the third convolution layer, and restore the 32 features to the dimensions of the input channel through the fourth convolution layer.

[0115] Additionally, the power generation prediction model (111) can apply a first fully connected layer including a ReLU (Rectified Linear Unit) activation function and a dropout layer to features extracted through a convolution layer, and output a final power generation prediction value through a second fully connected layer.

[0116] FIG. 6 is a diagram illustrating the operation of evaluating the predicted value of a power generation prediction model according to one embodiment of the present disclosure.

[0117] Referring to FIG. 6, the electronic device (100) can compare the amount of power generated predicted by the power generation prediction model (111) with the actual amount of power generated.

[0118] In one embodiment, the electronic device (100) can obtain the amount of power generated predicted for each of a plurality of time points through a power generation prediction model (111), and obtain the amount of power generated actually for each of the plurality of time points.

[0119] Accordingly, the electronic device (100) can compare the predicted amount of power generated with the actual amount of power generated through the power generation prediction model (111).

[0120] For example, the electronic device (100) may set a point in time when the difference between the power generation amount predicted by the power generation prediction model (111) and the actual power generation amount among a plurality of points in time exceeds a threshold value as a threshold point in time, and may calculate the time between the point in time when the power generation amount for a plurality of points in time is predicted by the power generation prediction model (111) and the threshold point in time and set it as a threshold time.

[0121] At this time, the electronic device (100) can set at least one of the points within a critical time as the target point when predicting the amount of power generated for the target point through the power generation prediction model (111).

[0122] At this time, when multiple threshold times are set, the electronic device (100) can calculate an average value for the multiple threshold times and set at least one of the points within the calculated average value from the point of prediction of power generation amount as a target point.

[0123] In addition, the electronic device (100) may set the target time based on user input.

[0124] FIG. 7 is a diagram illustrating the operation of an electronic device communicating with a manager device according to one embodiment of the present disclosure.

[0125] Referring to FIG. 7, the electronic device (100) can communicate with the manager device (200), and to do so, the electronic device (100) may include a communication interface.

[0126] A communication interface is a configuration for performing communication with the outside world.

[0127] The communication interface may include circuits, modules, chips, etc., for performing communication using various wired or wireless communication methods. The communication interface may also be connected to external devices and servers through various networks.

[0128] Depending on the area or scale, a network may be a Personal Area Network (PAN), Local Area Network (LAN), Wide Area Network (WAN), etc., and depending on the openness of the network, it may be an Intranet, Extranet, or Internet, etc.

[0129] The communication interface can be connected to external devices and servers through various wireless communication methods such as LTE (long-term evolution), LTE-A (LTE Advance), 5G (5th Generation) mobile communication, CDMA (code division multiple access), WCDMA (wideband CDMA), UMTS (universal mobile telecommunications system), WiBro (Wireless Broadband), GSM (Global System for Mobile Communications), DMA (Time Division Multiple Access), WiFi (Wi-Fi), WiFi Direct, Bluetooth, BLE (Bluetooth Low Energy), NFC (near field communication), Zigbee, and LoRa.

[0130] In addition, the communication interface may be connected to external devices and servers via wired communication methods such as Ethernet, optical networks, USB (Universal Serial Bus), and Thunderbolt.

[0131] In addition, the communication interface may be configured to utilize various newly devised communication methods / technologies in the future.

[0132] In one embodiment, the electronic device (100) can transmit prediction information including the predicted amount of power generation for a target time to a manager's device (200) that manages the power generation device.

[0133] In addition, the electronic device (100) may set the target time based on user input received from the manager device (200).

[0134] Meanwhile, the various embodiments described above may be implemented in a recording medium readable by a computer or a similar device using software, hardware, or a combination thereof.

[0135] According to hardware implementation, the embodiments described in this disclosure may be implemented using at least one of ASICs (Application Specific Integrated Circuits), DSPs (digital signal processors), DSPDs (digital signal processing devices), PLDs (programmable logic devices), FPGAs (field programmable gate arrays), processors, controllers, microcontrollers, microprocessors, and other electrical units for performing functions.

[0136] In some cases, the embodiments described herein may be implemented as the processor itself. In a software implementation, embodiments such as the procedures and functions described herein may be implemented as separate software modules. Each of the aforementioned software modules may perform one or more functions and operations described herein.

[0137] Meanwhile, computer instructions for performing processing operations in electronic devices, etc., according to the various embodiments of the present disclosure described above may be stored in a non-transitory computer-readable medium. When computer instructions stored in such a non-transitory computer-readable medium are executed by a processor of a specific device, they cause the specific device described above to perform processing operations according to the various embodiments described above.

[0138] A non-transient computer-readable medium refers to a medium that stores data semi-permanently and can be read by a device, unlike media that store data for a short period of time such as registers, caches, and memory. Specific examples of non-transient computer-readable media include CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, and ROMs.

[0139] Although preferred embodiments of the present disclosure have been illustrated and described above, the present disclosure is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the present disclosure as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present disclosure.

[0140] <Contents of the Discipline>

[0141] □ Korean

[0142] This project (result) is the outcome of the Regional Innovation-Centered University Support System (RISE), carried out with funding from the Ministry of Education and Gangwon Special Self-Governing Province and supported by the Gangwon RISE Center in 2025. (2025-RISE-10-002)

[0143]

[0144] The electronic device and the method of operation according to the present disclosure can be applied to various renewable energy generation facilities where changes in power generation over time occur, such as photovoltaic power generation devices, wind power generation devices, hydroelectric power generation devices, and geothermal power generation devices. In particular, since the power generation forecasting technology according to the present disclosure can predict with high precision by decomposing the short- and long-term variability of power generation into intra-period and inter-period patterns, it can provide real-time or reservation-based power generation forecasting information to power system operators, power generation business operators, energy storage device operators, etc., and can be usefully utilized for grid stabilization, power generation planning, peak load management, and demand response (DR) operation.

[0145] In addition, the electronic device according to the present disclosure collects power generation data from multiple power generation facilities through various information and communication networks and can be implemented in the form of a cloud server, an edge server, or a local server within an individual power plant; thus, it can be industrially utilized in various infrastructure environments, such as not only large-scale power complexes but also small-scale distributed power generation systems, microgrids, and self-generation systems for homes or buildings.

[0146] In addition, the power generation prediction model and the method of operation according to the present disclosure can be extended to energy-related value-added services, such as power demand forecasting, electric vehicle charging demand forecasting, and energy price forecasting, targeting time-series data similar in form to power generation data, and can be implemented in the form of software or hardware in conventional electronic devices such as servers, gateways, industrial computers, and cloud computing infrastructure.

Claims

1. In a method of operating an electronic device, A step of acquiring 1D power generation data including power generation amount over time; A step of obtaining 2D matrix data including at least one power generation cycle and an amplitude related to a power generation amount corresponding to the power generation cycle based on the above 1D power generation data; and A method of operating an electronic device comprising: a step of predicting the amount of power generated for at least one target time point based on feature information of the above 2D matrix data.

2. In Paragraph 1, The step of acquiring the above 2D matrix data is, A step of obtaining transformed data including frequency and amplitude related to power generation amount by performing a Fast Fourier Transform on the above 1D power generation data; and A method of operating an electronic device comprising the step of generating the 2D matrix data based on the above-mentioned transformation data.

3. In Paragraph 2, The step of generating the above 2D matrix data is, Identifying frequencies corresponding to each of a plurality of amplitudes having amplitude values ​​exceeding a reference value from the above-mentioned conversion data, and identifying a plurality of generation cycles based on the identified frequencies, A method of operation of an electronic device, wherein, among the identified plurality of power generation cycles, a pattern within the cycle indicating a phase difference between different points in time within the same power generation cycle is arranged in rows, and a pattern between the cycles indicating a time difference between points having the same phase within different power generation cycles is arranged in columns to generate the 2D matrix data.

4. In Paragraph 3, The step of predicting the amount of power generated for at least one target time point is, A step of applying at least one convolution layer to the above 2D matrix data to extract feature information related to the in-period pattern and the period-period pattern; A step of converting the above-mentioned extracted feature information into a one-dimensional form; and A method of operating an electronic device comprising: a step of predicting the amount of power generated for the target time point based on feature information converted into the above one-dimensional form.

5. In Paragraph 4, The step of converting the above-mentioned extracted feature information into a one-dimensional form is, A method of operation of an electronic device for converting into a one-dimensional form by connecting patterns corresponding to each of at least one row and at least one column constituting the above-mentioned extracted feature information.

6. In Paragraph 5, The step of predicting the amount of power generated for the target time point based on the feature information converted into the above one-dimensional form is: Based on the feature information converted into the above one-dimensional form and the above 2D matrix data, intra-cycle changes and inter-cycle changes for the above plurality of power generation cycles are identified, and A method of operation of an electronic device for obtaining a predicted amount of power generation for a target time point according to a power generation cycle corresponding to the target time point, based on the changes within the cycle and changes during the cycle identified above.

7. In Paragraph 1, The method of operation of the above electronic device is, A step of obtaining a predicted power generation history including a predicted power generation amount for each of a plurality of time points and a power generation history including an actual power generation amount actually generated for each of the plurality of time points; A step of identifying a point in time when the difference between the predicted power generation amount and the actual power generation amount exceeds a threshold value based on the predicted power generation history and the power generation history; A step of calculating the time between the time when the power generation amount for the identified time is predicted and the identified time and setting it as a threshold time; and A method of operating an electronic device comprising the step of setting at least one of the points within the threshold time from the point of predicting the amount of power generated as the target point.

8. In electronic devices, Memory storing a power generation prediction model for predicting power generation over time; and An electronic device comprising: a processor connected to the memory and inputting 1D power generation data including power generation amount over time into the power generation prediction model to obtain a power generation amount for at least one target time point based on the output data.

9. In Paragraph 8, The above processor is, Through the above power generation prediction model, a Fast Fourier Transform is performed on the above 1D power generation data to obtain transformed data including frequency and amplitude related to power generation, and An electronic device that, through the power generation prediction model, identifies a frequency corresponding to each of a plurality of amplitudes having an amplitude value exceeding a reference value from the converted data, identifies a plurality of power generation cycles based on the identified frequency, arranges a cycle pattern representing a phase difference between different points in time within the same power generation cycle among the identified plurality of power generation cycles in rows, and arranges a cycle pattern representing a temporal difference between points in time having the same phase within different power generation cycles in columns to obtain 2D matrix data.

10. In Paragraph 9, The above processor is, Through the above power generation prediction model, at least one convolution layer is applied to the above 2D matrix data to extract feature information related to two-dimensional changes over time, and the feature information is converted into a one-dimensional form by connecting patterns corresponding to each of at least one row and at least one column constituting the extracted feature information. An electronic device that obtains the amount of power generated for the target time point predicted based on the feature information converted into the one-dimensional form and the 2D matrix data through the power generation prediction model.

11. A non-transient computer-readable medium storing at least one instruction that is executed by a processor of an electronic device to cause the electronic device to perform the method of operation of claim 1.

Images