Energy Internet energy load prediction method, device and electronic equipment

Abstract

The present invention provides a method, device, and electronic device for predicting energy load in an energy internet, relating to the technical field of energy internet. The method comprises: obtaining a noisy load data sequence for a target area within the energy internet within a preset time period; and processing the noisy load data sequence using a noise-free load amplitude prediction model to obtain the amplitude of the noise-free energy load data for the target area at the next moment. The present invention enables accurate prediction of noise-free load data in the energy internet.

Description

Technical Field

[0001] The present invention relates to the technical field of energy internet, and in particular to a method, device and electronic equipment for predicting energy load of energy internet. Background Art

[0002] As an emerging regional energy system, the Energy Internet's efficient and robust operation relies heavily on advanced artificial intelligence technologies such as neural networks. To achieve efficient operation of the Energy Internet, timely and accurate load forecasting is crucial.

[0003] However, during the actual operation of the Energy Internet, collected load data often contains a certain degree of noise due to internal system factors and external security attacks. This noise can interfere with the normal operation of the Energy Internet, affect the accuracy of system modeling and optimization, and in extreme cases, even cause system failure or paralysis. Estimating the noise level of collected load data requires overcoming the challenges posed by the noise component and the uncertainty and randomness of the original load data. Effective solutions are currently lacking in the relevant technologies.

[0004] Therefore, how to achieve accurate prediction of noise-free load data of the Energy Internet is a technical problem that needs to be solved urgently. Summary of the Invention

[0005] In response to the above-mentioned problems existing in the prior art, the present invention provides a method, device and electronic equipment for predicting energy load of an energy internet, so as to realize accurate prediction of noise-free load data of the energy internet.

[0006] The present invention provides a method for predicting energy load of an energy internet, comprising the following steps.

[0007] A noisy load data sequence of a target area in an energy internet within a preset time period is obtained; the noisy load data sequence is processed using a noise-free load amplitude prediction model to obtain the amplitude of the noise-free energy load data of the target area at the next moment; wherein the noise-free load amplitude prediction model includes a denoising autoencoder network and an amplitude estimation network, the denoising autoencoder network is used to process the noisy load data sequence to obtain a noise-free load data sequence, and the amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment.

[0008] According to a method for predicting energy load of an energy internet provided by the present invention, the denoising autoencoder network includes an encoder implemented using a ReLU function and a decoder implemented using a sigmoid function; the noise-free load data sequence is obtained by using the denoising autoencoder network in the following manner: the noisy load data sequence is processed using a ReLU function to obtain a dimensionality-reduced load data representation; based on the dimensionality-reduced load data representation, data is reconstructed using a sigmoid function to obtain the noise-free load data sequence.

[0009] According to a method for predicting energy load of an energy internet provided by the present invention, the amplitude estimation network is a multi-layer backward propagation network; the amplitude estimation network is used to obtain the amplitude of the noise-free energy load data of the target area at the next moment in the following manner: the noise-free load data sequence is input into the multi-layer backward propagation network to obtain the amplitude of the noise-free energy load data of the target area at the next moment output by the multi-layer backward propagation network.

[0010] According to a method for predicting energy load of an energy internet provided by the present invention, the method further includes: repeatedly executing the steps of obtaining a noisy load data sequence of the energy internet, and processing the noisy load data sequence using a noise-free load amplitude prediction model, until the number of noisy load data sequences obtained reaches a sampling threshold; for each noisy load data sequence, determining the noise level of the noisy load data sequence according to the difference between the amplitude of the noisy load data sequence and the amplitude of the noise-free load data sequence corresponding to the noisy load data sequence; in response to the noise levels of the obtained noisy load data sequences being greater than a preset noise level threshold, determining that the load data sequence of the target area is subject to noise interference or signal attack during transmission.

[0011] According to a method for predicting energy load of an energy internet provided by the present invention, the sampling threshold is determined in the following manner: a single-step estimation probability of a prediction error is obtained based on the prediction accuracy of the noise-free load amplitude prediction model; wherein the prediction accuracy of the noise-free load amplitude prediction model is determined based on the difference between the noise level of a noise-carrying load data sequence obtained using the noise-free load amplitude prediction model and the actual noise level of the noise-carrying load data sequence; the sampling threshold is obtained based on the single-step estimation probability and a preset probability threshold.

[0012] According to a method for predicting energy load of an energy internet provided by the present invention, the noise-free load amplitude prediction model is trained in the following manner: a first training data set, a second training data set and a third training data set are obtained; wherein, the first training data set includes a noise-carrying historical load data sequence as a sample and a noise-free historical load data sequence as a label, the second training data set includes a noise-free historical load data sequence as a sample and the amplitude of the load data at the next moment of the noise-free historical load data sequence as a label, and the third training data set includes a noise-carrying historical load data sequence as a sample and the amplitude of the load data at the next moment of the noise-free historical load data sequence as a label; the first training data set is used to train an initial denoising autoencoder network to obtain a trained denoising autoencoder network; the second training data set is used to train an initial amplitude estimation network to obtain a trained amplitude estimation network; the third training data set is used to jointly train the trained denoising autoencoder network and the trained amplitude estimation network to obtain the trained noise-free load amplitude prediction model.

[0013] The present invention also provides a device for predicting energy load of an energy internet, comprising the following modules: a first acquisition module, for acquiring a noise-carrying load data sequence of a target area in the energy internet within a preset time period; a second acquisition module, for processing the noise-carrying load data sequence using a noise-free load amplitude prediction model to obtain the amplitude of the noise-free energy load data of the target area at the next moment; wherein the noise-free load amplitude prediction model comprises a denoising autoencoder network and an amplitude estimation network, the denoising autoencoder network is used to process the noise-carrying load data sequence to obtain a noise-free load data sequence, and the amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for predicting energy load of the energy internet as described above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for predicting energy loads of the energy internet.

[0016] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described methods for predicting energy loads of the energy internet.

[0017] The energy load prediction method, device, and electronic device provided by this invention utilize a noise-free load amplitude prediction model to process a noisy load data sequence for a target area within the energy internet within a preset time period, and obtain the noise-free energy load data amplitude for the target area at the next moment. Leveraging the powerful data processing and data feature learning capabilities of a machine learning model, this method effectively overcomes the effects of noise in the load data sequence, as well as the inherent uncertainty and randomness of the load data sequence, on the accuracy of load data amplitude prediction, thereby enabling accurate prediction of noise-free load data for the energy internet. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is one of the flow charts of the energy load prediction method of the energy internet provided by the present invention.

[0020] Figure 2 This is the second flow chart of the method for predicting energy load of the energy internet provided by the present invention.

[0021] Figure 3 It is a flow chart of a method provided by the present invention for determining whether a load data sequence of a target area is subject to noise interference or signal attack during transmission.

[0022] Figure 4 It is a flow chart of the training method of the noise-free load amplitude prediction model provided by the present invention.

[0023] Figure 5 It is a structural diagram of the noise reduction autoencoding network provided by the present invention.

[0024] Figure 6 It is a schematic diagram of the structure of the multi-layer back-propagation network provided by the present invention.

[0025] Figure 7 It is a schematic diagram of the structure of the residual network provided by the present invention.

[0026] Figure 8 It is a structural schematic diagram of the energy load prediction device of the energy internet provided by the present invention.

[0027] Figure 9 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0028] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0029] The following combination Figure 1-Figure 7 The present invention describes a method for predicting energy load of an energy internet.

[0030] Figure 1 This is one of the flow charts of the energy load prediction method of the energy internet provided by the present invention, which is executed by the control system of the energy internet, such as Figure 1 As shown, the method includes the following:

[0031] Step 101: Obtain a load data sequence carrying noise in a target area in the energy internet within a preset time period.

[0032] The load data series for a target area refers to a series of data points that record the time-varying changes in the power load in the target area of ​​the Energy Internet. These data points are usually arranged in chronological order and can reflect the fluctuations in the power demand of the Energy Internet over different time periods.

[0033] The preset time period refers to a time period of a certain length, such as 15 minutes or 30 minutes before the current moment. In a specific implementation, the selection of the prediction time period can be determined based on the processing length of the input data by the noise-free load amplitude prediction model (for a detailed description, refer to the relevant content in step 102, which will not be repeated here).

[0034] The load data of the Energy Internet is usually collected through measuring equipment (such as smart meters, sensors, etc.) installed at key nodes of the Energy Internet, and transmitted to the control system for analysis and processing through various methods such as power line carrier, wireless public network, and fiber optic communication.

[0035] Step 102: Use the noise-free load amplitude prediction model to process the noisy load data sequence to obtain the amplitude of the noise-free energy load data of the target area at the next moment.

[0036] The noise-free load amplitude prediction model includes a denoising autoencoder network and an amplitude estimation network. The denoising autoencoder network is used to process the noisy load data sequence to obtain a noise-free load data sequence. The amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment.

[0037] In the specific implementation process, Figure 5 As shown, the obtained noisy load data sequence can be input into the noise-free load amplitude prediction model; the noise-free load amplitude prediction model obtained for different training data can output the predicted amplitude of the load data at the next moment, or can output the predicted amplitude sequence of the load data sequence for the next time period (a sequence composed of predicted amplitudes at multiple time points), which is not limited to the description of this specification.

[0038] In the specific implementation process, a denoising autoencoder network can be constructed in a variety of ways and is not limited to the description in this specification.

[0039] In a specific implementation process, the amplitude estimation network can be constructed in a variety of ways, which are not limited to the description in this specification.

[0040] For an embodiment of the denoising autoencoder network and the amplitude estimation network, see Figure 2 The relevant content in will not be repeated here.

[0041] For the training method of the noiseless load amplitude prediction model, see Figure 4 The relevant content in will not be repeated here.

[0042] Figure 2 This is the second flow chart of the method for predicting energy load of energy internet provided by the present invention. In this embodiment, the noise reduction autoencoder network includes an encoder implemented by ReLU function and a decoder implemented by sigmoid function, and the amplitude estimation network is a multi-layer back propagation network. Figure 2 As shown, the method includes the following:

[0043] Step 201: Obtain a load data sequence carrying noise of the energy internet.

[0044] For a detailed description of this step, please refer to the relevant content in step 101, which will not be repeated here.

[0045] Step 202: Use the ReLU function to process the noisy load data sequence to obtain a dimensionality-reduced load data representation.

[0046] The ReLU function is as follows:

[0047] (1)

[0048] in, For input data.

[0049] In the specific implementation process, the load data sequence carrying noise can be used as the input data of the ReLU function in turn. , after dimensionality reduction processing as shown in formula (1), the load data representation is obtained.

[0050] Step 203: Based on the load data representation after dimensionality reduction, the sigmoid function is used to reconstruct the data to obtain a noise-free load data sequence.

[0051] The sigmoid function is shown below

[0052] (2)

[0053] in, For input data.

[0054] In the specific implementation process, the load data obtained in step 202 can be expressed as the input data of the sigmoid function , after data reconstruction as shown in formula (2), a noise-free load data sequence is obtained.

[0055] Step 204: Input the noise-free load data sequence into the multi-layer backward propagation network to obtain the amplitude of the noise-free energy load data of the target area at the next moment output by the multi-layer backward propagation network.

[0056] like Figure 6 As shown, the multi-layer back propagation network includes an input layer, a hidden layer and an output layer. In some embodiments, the back propagation network can be Figure 7 The residual network shown improves the estimation performance of multi-layer backpropagation networks.

[0057] Figure 3 FIG. 1 is a flow chart of a method for determining whether a load data sequence of a target area is subject to noise interference or signal attack during transmission provided by the present invention. Figure 3 As shown, the method includes the following:

[0058] Step 301: Acquire a load data sequence carrying noise whose number reaches a sampling threshold.

[0059] In a specific implementation process, steps 101 to 102 may be repeatedly performed to obtain a plurality of consecutive noisy load data sequences and obtain the amplitude of the corresponding noise-free energy load data until the number of acquired sequences reaches a sampling threshold.

[0060] In some embodiments, the sampling threshold may be determined by:

[0061] According to the prediction accuracy of the noise-free load amplitude prediction model, the single-step estimation probability of prediction error is obtained; according to the single-step estimation probability and the preset probability threshold, the sampling threshold is obtained.

[0062] The prediction accuracy of the noise-free load amplitude prediction model is determined by the difference between the noise level of the noisy load data sequence obtained using the noise-free load amplitude prediction model and the actual noise level of the noisy load data sequence (which can be obtained from actual instrument measurement or simulation calculation). The calculation formula is as follows:

[0063] (3)

[0064] is the actual noise level, is the noise level of the noisy load data sequence obtained by the noise-free load amplitude prediction model, N is the number of sampling calculations, and level is the number of noise levels; is the prediction accuracy of the noise-free load amplitude prediction model.

[0065] One-step probability estimation The calculation formula is as follows:

[0066] (4)

[0067] in, is the prediction accuracy of the noise-free load amplitude prediction model, that is, .

[0068] As an example only, setting the preset probability threshold to a, the following formula can be used to estimate the probability based on the single step , and the preset probability threshold, the sampling threshold M is obtained.

[0069] (5)

[0070] It can be seen from formula (5) that if more than M samples are judged to have a noise level greater than 1, the preset probability threshold a (for example, 0.99) can be used to determine that there is noise interference or signal attack behavior in the load data collection and transmission process of the energy Internet, thereby determining the size of M.

[0071] Step 302: For each noisy load data sequence, determine the noise level of the noisy load data sequence according to the difference between the amplitude of the noisy load data sequence and the amplitude of the noise-free load data sequence corresponding to the noisy load data sequence.

[0072] During implementation, the noise level of a noisy load data sequence can be determined in a variety of ways, not limited to the description herein. For example, an adaptive classification algorithm (e.g., a k-means clustering algorithm) can be used to determine the noise level of the noisy load data sequence based on the difference between the amplitude of the noisy load data sequence and the amplitude of the noise-free load data sequence corresponding to the noisy load data sequence. Another example is noise threshold level quantization, where the noise level of the noisy load data sequence is determined based on the difference between the amplitude of the noisy load data sequence and the amplitude of the noise-free load data sequence corresponding to the noisy load data sequence.

[0073] Step 303: In response to the noise levels of the acquired load data sequences carrying noise being greater than a preset noise level threshold, it is determined that the load data sequences in the target area are subject to noise interference or signal attack during transmission.

[0074] As an example only, assume that the noise level is 5, and the noise is always positive, and the noise level 1 is a noise-free situation. If the noise levels of the acquired load data sequences carrying noise are all greater than the preset noise level threshold of level 1, it can be considered that the load data sequence in the target area is subject to noise interference or signal attack during the transmission process, and corresponding countermeasures can be taken, such as fault warning, safety warning, fault troubleshooting and other measures.

[0075] In the embodiment provided by the present invention, for each noisy load data sequence obtained, the noise level of the noisy load data sequence is determined based on the difference between the amplitude of the noisy load data sequence and the amplitude of the noise-free load data sequence corresponding to the noisy load data sequence; when the noise levels of the obtained noisy load data sequences are all greater than the preset noise level threshold, the load data sequence of the target area can be determined in a timely and reliable manner, and noise interference or signal attack exists during the transmission process, and countermeasures can be taken in a timely manner to improve the modeling and optimization performance of the energy Internet, thereby ensuring the stability and robustness of the operation of the energy Internet.

[0076] Figure 4 is a flow chart of the training method of the noiseless load amplitude prediction model provided by the present invention, such as Figure 4 As shown, the method includes the following:

[0077] Step 401: Obtain a first training data set, a second training data set, and a third training data set.

[0078] The first training data set includes a historical load data sequence with noise as a sample and a historical load data sequence without noise as a label.

[0079] In a specific implementation, the first training dataset can be obtained in a variety of ways, not limited to the description in this specification. For example, an original load data sequence can be obtained, random noise can be added to the original load data sequence to obtain a historical load data sequence with noise as sample data, and the original load data sequence can be used as the label of the sample data.

[0080] The second training data set includes a noise-free historical load data sequence as a sample and an amplitude of load data at the next moment of the noise-free historical load data sequence as a label.

[0081] During the specific implementation process, the second training data set can be obtained in a variety of ways, not limited by the description of this specification. For example, the original load data sequence can be divided into two time periods: the first time period and the second time period. The load data of the first time period is used as the sample data, and the amplitude of the load data of the second time period is used as the label of the sample data. If the second time period includes multiple load data, the noise-free load amplitude prediction model obtained by training can output the amplitudes of multiple load data in the next time period; if the second time period includes one load data, the noise-free load amplitude prediction model obtained by training can output the amplitude of the load data at the next moment.

[0082] The third training data set includes a historical load data sequence with noise as a sample and an amplitude of load data at the next moment of the historical load data sequence without noise as a label.

[0083] During the specific implementation process, the third training data set can be obtained in a variety of ways, not limited by the description in this specification. For example, the historical load data sequence with noise can be divided into two time periods: the first time period and the second time period. The load data with noise in the first time period is used as sample data, and the amplitude of the noise-free load data in the second time period is used as the label of the sample data. If the second time period includes multiple load data, the noise-free load amplitude prediction model obtained by training can output the amplitudes of multiple load data in the next time period; if the second time period includes one load data, the noise-free load amplitude prediction model obtained by training can output the amplitude of the load data at the next moment.

[0084] In the specific implementation process, by collecting historical load data carrying noise and considering its collection density and frequency, the load forecast of the energy Internet in different time periods and different area sizes can be achieved.

[0085] Step 402: Use the first training data set to train the initial denoising autoencoder network to obtain a trained denoising autoencoder network.

[0086] In a specific implementation, the denoising autoencoder network can be trained using the first training dataset using a layer-by-layer greedy training approach. When training a layer, the weight parameters of that layer are first randomly initialized. A restoration layer is then added after that layer. The weight parameters of that layer are updated by minimizing the mean squared error between the restored signal and the original signal. The restoration layer is then removed, and training proceeds to the next layer. This method continues layer by layer until a training termination condition, such as model convergence, is met, resulting in the trained denoising autoencoder network.

[0087] Step 403: Use the second training data set to train the initial amplitude estimation network to obtain a trained amplitude estimation network.

[0088] When training a multi-layer backpropagation network, a second training dataset can be used, with the tanh function or ReLU function as the activation function, combined with an appropriate loss function (e.g., mean squared error) and optimization algorithm (e.g., gradient descent). During training, the network updates its weight parameters using the backpropagation algorithm based on the gradient of the loss function to minimize the loss function until the training end condition, such as model convergence, is met, resulting in a trained amplitude estimation network.

[0089] The tanh function is shown below:

[0090] (6)

[0091] in, x For input data.

[0092] The Relu function is as follows:

[0093] (7)

[0094] in, x For input data.

[0095] Step 404: Using the third training data set, jointly train the trained denoising autoencoder network and the trained amplitude estimation network to obtain a trained noise-free load amplitude prediction model.

[0096] In the specific implementation process, in order to obtain higher prediction accuracy, the third training data set can be used to connect the trained denoising autoencoder network and the trained amplitude estimation network in series and then perform joint training.

[0097] In the embodiment provided by the present invention, based on the periodic characteristics (explicit or implicit) of the load data of the energy Internet and the high correlation between the current noise distribution and the historical noise distribution, a noise-carrying historical load data sequence and a noise-free historical load data sequence are used to train a noise-free load amplitude prediction model that can accurately predict the load data amplitude of the noisy load data sequence at the next moment. The noise-free load amplitude prediction model can then be used to timely and accurately predict the operating load of the energy Internet.

[0098] The following describes the energy load prediction device for the energy internet provided by the present invention. The energy load prediction device for the energy internet described below and the energy load prediction method for the energy internet described above can refer to each other.

[0099] Figure 8 This is a schematic diagram of the structure of the energy load prediction device of the energy internet provided by the present invention. Figure 8 As shown, the apparatus 800 includes the following modules.

[0100] The first acquisition module 810 is configured to acquire a load data sequence carrying noise in a target area of ​​the energy internet within a preset time period.

[0101] The second acquisition module 820 is configured to process the noisy load data sequence using a noise-free load amplitude prediction model to obtain the amplitude of the noise-free energy load data of the target area at the next moment.

[0102] Among them, the noise-free load amplitude prediction model includes a denoising autoencoder network and an amplitude estimation network. The denoising autoencoder network is used to process the noisy load data sequence to obtain a noise-free load data sequence, and the amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment.

[0103] Figure 9 An example of a physical structure diagram of an electronic device is shown below. Figure 9As shown, the electronic device may include: a processor 910, a communication interface 920, a memory 930, and a communication bus 940. The processor 910, the communication interface 920, and the memory 930 communicate with each other via the communication bus 940. The processor 910 may invoke logic instructions in the memory 930 to execute a method for predicting energy load in the energy internet. The method includes: obtaining a noisy load data sequence for a target area in the energy internet within a preset time period; processing the noisy load data sequence using a noise-free load amplitude prediction model to obtain the amplitude of the noise-free energy load data for the target area at the next moment; wherein the noise-free load amplitude prediction model includes a denoising autoencoder network and an amplitude estimation network. The denoising autoencoder network is used to process the noisy load data sequence to obtain a noise-free load data sequence, and the amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment.

[0104] Furthermore, the logic instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0105] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the energy Internet energy load prediction method provided by the above methods, which includes: obtaining a noise-carrying load data sequence of a target area in the energy Internet within a preset time period; using a noise-free load amplitude prediction model to process the noise-carrying load data sequence to obtain the amplitude of the noise-free energy load data of the target area at the next moment; wherein the noise-free load amplitude prediction model includes a denoising autoencoder network and an amplitude estimation network, the denoising autoencoder network is used to process the noise-carrying load data sequence to obtain a noise-free load data sequence, and the amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment.

[0106] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the energy internet energy load prediction method provided by the above-mentioned methods, the method comprising: obtaining a noise-carrying load data sequence of a target area in the energy internet within a preset time period; using a noise-free load amplitude prediction model to process the noise-carrying load data sequence to obtain the amplitude of the noise-free energy load data of the target area at the next moment; wherein the noise-free load amplitude prediction model comprises a denoising autoencoder network and an amplitude estimation network, the denoising autoencoder network is used to process the noise-carrying load data sequence to obtain a noise-free load data sequence, and the amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment.

[0107] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0108] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for predicting energy load of energy internet, characterized in that: include: Obtaining a load data sequence with noise in a target area of ​​the energy internet within a preset time period; Using a noise-free load amplitude prediction model, the load data sequence carrying noise is processed to obtain the amplitude of the noise-free energy load data of the target area at the next moment; The noise-free load amplitude prediction model includes a denoising autoencoder network and an amplitude estimation network. The denoising autoencoder network is used to process the noisy load data sequence to obtain a noise-free load data sequence. The amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment. The denoising autoencoder network includes an encoder implemented by a ReLU function and a decoder implemented by a sigmoid function. The noise-free load data sequence is obtained by using the denoising autoencoder network in the following manner: Processing the noisy load data sequence using a ReLU function to obtain a dimensionality-reduced load data representation; Based on the load data representation after dimensionality reduction, data reconstruction is performed using a sigmoid function to obtain the noise-free load data sequence; The amplitude estimation network is a multi-layer back-propagation network. The amplitude estimation network is used to obtain the noise-free energy load data amplitude of the target area at the next moment in the following manner: The noise-free load data sequence is input into the multi-layer backward propagation network to obtain the amplitude of the noise-free energy load data of the target area at the next moment output by the multi-layer backward propagation network.

2. The method for predicting energy load of energy internet according to claim 1, characterized in that: The method further comprises: Repeating the steps of obtaining a noisy load data sequence of the energy internet and processing the noisy load data sequence using a noise-free load amplitude prediction model until the number of obtained noisy load data sequences reaches a sampling threshold; For each of the noisy load data sequences, determining a noise level of the noisy load data sequence according to a difference between an amplitude of the noisy load data sequence and an amplitude of the noise-free load data sequence corresponding to the noisy load data sequence; In response to the noise levels of the acquired load data sequences carrying noise being greater than a preset noise level threshold, it is determined that the load data sequences in the target area are subject to noise interference or signal attack during transmission.

3. The method for predicting energy load of energy internet according to claim 2, characterized in that: The sampling threshold is determined in the following manner: Obtaining a single-step estimated probability of prediction error based on the prediction accuracy of the noise-free load amplitude prediction model; wherein the prediction accuracy of the noise-free load amplitude prediction model is determined based on a difference between a noise level of a noisy load data sequence obtained using the noise-free load amplitude prediction model and an actual noise level of the noisy load data sequence; The sampling threshold is obtained according to the single-step estimation probability and a preset probability threshold.

4. The method for predicting energy load of energy internet according to claim 1, characterized in that: The noise-free load amplitude prediction model is trained using the following method: Obtain a first training data set, a second training data set, and a third training data set; wherein the first training data set includes a historical load data sequence with noise as a sample and a historical load data sequence without noise as a label; the second training data set includes a historical load data sequence with noise as a sample and an amplitude of load data at a next moment of the historical load data sequence without noise as a label; and the third training data set includes a historical load data sequence with noise as a sample and an amplitude of load data at a next moment of the historical load data sequence without noise as a label; Using the first training data set, training the initial denoising autoencoder network to obtain a trained denoising autoencoder network; Using the second training data set, training the initial amplitude estimation network to obtain a trained amplitude estimation network; The trained denoising autoencoder network and the trained amplitude estimation network are jointly trained using the third training data set to obtain the trained noise-free load amplitude prediction model.

5. A device for predicting energy load of energy internet, characterized in that: include: The first acquisition module is used to obtain a load data sequence carrying noise in a target area in the energy internet within a preset time period; a second acquisition module, configured to process the noisy load data sequence using a noise-free load amplitude prediction model to obtain the amplitude of the noise-free energy load data of the target area at the next moment; The noise-free load amplitude prediction model includes a denoising autoencoder network and an amplitude estimation network. The denoising autoencoder network is used to process the noisy load data sequence to obtain a noise-free load data sequence. The amplitude estimation network is used to process the noise-free load data sequence to obtain the amplitude of the noise-free energy load data at the next moment. The denoising autoencoder network includes an encoder implemented by a ReLU function and a decoder implemented by a sigmoid function. The noise-free load data sequence is obtained by using the denoising autoencoder network in the following manner: Processing the noisy load data sequence using a ReLU function to obtain a dimensionality-reduced load data representation; Based on the load data representation after dimensionality reduction, data reconstruction is performed using a sigmoid function to obtain the noise-free load data sequence; The amplitude estimation network is a multi-layer back-propagation network. The amplitude estimation network is used to obtain the noise-free energy load data amplitude of the target area at the next moment in the following manner: The noise-free load data sequence is input into the multi-layer backward propagation network to obtain the amplitude of the noise-free energy load data of the target area at the next moment output by the multi-layer backward propagation network.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for predicting energy load of the energy internet according to any one of claims 1 to 4 is implemented.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting energy load of the energy internet as claimed in any one of claims 1 to 4 is implemented.

8. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting energy load of the energy internet as claimed in any one of claims 1 to 4 is implemented.

Images