Electromagnetic situation prediction method and device based on Informer model

Abstract

The invention provides an electromagnetic situation prediction method and device based on an Informer model. The method comprises the following steps: acquiring electromagnetic signal data of a target area; performing data division processing on the electromagnetic signal data based on a time sequence to obtain a plurality of electromagnetic signal subsequences with fixed lengths; inputting the plurality of electromagnetic signal subsequences into a current pre-trained Informer model for situation prediction processing to obtain an electromagnetic situation prediction result; the current pre-trained Informer model is obtained based on a sparse attention mechanism, self-attention distillation and a generative decoder mechanism by adopting incremental learning training of electromagnetic signal sample data. According to the method, the Informer model is introduced, multi-scale analysis and time sequence modeling can be carried out on electromagnetic signal data in an area, accurate prediction of the electromagnetic situation is achieved, in addition, incremental learning is further introduced on the basis of a prediction framework, and the adaptability and flexibility of the prediction framework are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic situation prediction, and in particular to an electromagnetic situation prediction method and device based on an Informer model. Background Art

[0002] When performing security tasks for major events in key areas, the electromagnetic situation and its changes are often very complex. In order to ensure the smooth progress of the task, the electromagnetic situation must be accurately evaluated and predicted, which is a key prerequisite for achieving frequency safety. In this process, the generation and analysis of the electromagnetic situation requires not only mastering the current usage status of the spectrum space, but also in-depth prediction of the comprehensive situation and future development trends of the spectrum space. This analysis not only focuses on the current use of the spectrum, but also pays more attention to revealing the role of spectrum evolution laws in the electromagnetic space situation, thereby providing support for the global prediction and dynamic evaluation of the electromagnetic situation.

[0003] The prediction and deduction of electromagnetic situation is to add the time dimension to the existing situation to form a multi-dimensional time series. This prediction method accurately predicts the changes in the electromagnetic environment through time series analysis and adaptive training, and then obtains the prediction results of the electromagnetic situation. This not only helps to intuitively display the electromagnetic situation, but also provides a basis for decision-making evaluation.

[0004] The existing electromagnetic situation prediction technologies are mainly based on time series analysis. These methods use historical data to train models and predict future spectrum usage. For example, autoregressive models (AR), moving average models (MA), and other statistical methods based on time series data have been applied in some communication systems. These methods can predict changes in electromagnetic situation to a certain extent and provide support for dynamic allocation of spectrum. However, these traditional methods often show low efficiency when faced with high-dimensional, complex, and multi-source electromagnetic data. They are difficult to cope with complex electromagnetic environments and dynamic changes in spectrum resources. They often lack sufficient flexibility when processing multidimensional data and cannot accurately capture the real-time dynamic changes of spectrum resources, which leads to poor accuracy in electromagnetic situation prediction. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides an electromagnetic situation prediction method and device based on an Informer model.

[0006] The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0007] In a first aspect, the present invention provides an electromagnetic situation prediction method based on an Informer model, comprising:

[0008] Acquire electromagnetic signal data of the target area;

[0009] The electromagnetic signal data is divided and processed based on the time sequence to obtain a plurality of electromagnetic signal subsequences of fixed length;

[0010] Multiple electromagnetic signal subsequences are input into the current pre-trained Informer model for situation prediction processing to obtain the electromagnetic situation prediction result; the current pre-trained Informer model is based on sparse attention mechanism, self-attention distillation, and generative decoder mechanism, and is trained by incremental learning of electromagnetic signal sample data.

[0011] Optionally, the electromagnetic signal data is divided and processed based on the time sequence to obtain a plurality of electromagnetic signal subsequences of fixed length, including:

[0012] Sequentially dividing the electromagnetic signal data according to time to obtain a plurality of electromagnetic signal subsequences;

[0013] Normalizing the multiple electromagnetic signal subsequences to obtain multiple normalized subsequence data;

[0014] A cyclic filling method is used to perform sequence filling processing on multiple normalized subsequence data to obtain multiple electromagnetic signal subsequences with fixed lengths.

[0015] Optionally, the current pre-trained Informer model includes: an input layer, an embedding layer, a position encoding layer, an encoder module, a self-attention distillation module, a decoder module, and an output layer connected in sequence;

[0016] The encoder module is provided with: a local attention module, a first feedforward neural network module, and a first residual connection and layer normalization module connected in sequence;

[0017] The decoder module is provided with: a self-attention module, an encoder-decoder attention module, a second feedforward neural network module, and a second residual connection and layer normalization module connected in sequence;

[0018] Among them, the local attention module is used for the setting of sparse attention mechanism, and the encoder-decoder attention module is used for the setting of generative decoder mechanism.

[0019] Optionally, the training process of the pre-trained Informer model includes:

[0020] Acquire electromagnetic signal sample data of the target area;

[0021] The electromagnetic signal sample data is divided and processed based on the time sequence to obtain a plurality of electromagnetic signal sample subsequences of fixed length;

[0022] Input multiple electromagnetic signal sample subsequences into the Informer model, and continuously train the Informer model in the direction of decreasing loss function;

[0023] The Informer model that reaches the preset stop condition is used as the pre-trained Informer model.

[0024] Optionally, the Informer model that reaches a preset stop condition is used as a pre-trained Informer model, including:

[0025] The Informer model corresponding to the number of iterations greater than the iteration threshold is used as the pre-trained Informer model; or,

[0026] The Informer model corresponding to the loss function value being less than the preset loss threshold is used as the pre-trained Informer model.

[0027] Optionally, the electromagnetic situation prediction method based on the Informer model further includes:

[0028] The data fine-tuning method is used to perform incremental learning on the pre-trained Informer model.

[0029] Optionally, a data fine-tuning method is used to perform incremental learning processing on the pre-trained Informer model, including:

[0030] Acquire new electromagnetic signal sample data;

[0031] The newly added electromagnetic signal sample data is divided and processed based on the time sequence to obtain a plurality of newly added electromagnetic signal sample subsequences of fixed length;

[0032] Freeze the parameters of the preset modules in the pre-trained Informer model to obtain a frozen Informer model;

[0033] Multiple newly added sample subsequences of electromagnetic signals are input into the frozen Informer model for incremental learning and training to obtain the current pre-trained Informer model.

[0034] In a second aspect, the present invention provides an electromagnetic situation prediction device based on an Informer model, the electromagnetic situation prediction device based on the Informer model comprises: an acquisition unit, a division unit and a training unit;

[0035] The acquisition unit is used to: acquire electromagnetic signal data of the target area;

[0036] The division unit is used to: perform data division processing on the electromagnetic signal data based on the time sequence to obtain a plurality of electromagnetic signal subsequences of fixed length;

[0037] The training unit is used to: input multiple electromagnetic signal subsequences into the current pre-trained Informer model for situation prediction processing to obtain electromagnetic situation prediction results; the current pre-trained Informer model is based on sparse attention mechanism, self-attention distillation, and generative decoder mechanism, and is trained by incremental learning of electromagnetic signal sample data.

[0038] In a third aspect, the present invention provides an electromagnetic situation prediction device based on an Informer model, comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor. When the electromagnetic situation prediction device based on the Informer model is running, the processor and the storage medium communicate through the bus, and the processor executes the machine-readable instructions to perform the steps of the electromagnetic situation prediction method based on the Informer model in the first aspect as described above.

[0039] The present invention provides an electromagnetic situation prediction method and device based on an Informer model. The electromagnetic situation prediction method based on the Informer model includes: obtaining electromagnetic signal data of a target area; performing data division processing on the electromagnetic signal data based on time sequence to obtain multiple electromagnetic signal subsequences of fixed length; inputting multiple electromagnetic signal subsequences into a current pre-trained Informer model for situation prediction processing to obtain an electromagnetic situation prediction result; the current pre-trained Informer model is based on a sparse attention mechanism, self-attention distillation, and a generative decoder mechanism, and is obtained by incremental learning and training of electromagnetic signal sample data. In the present invention, by introducing the Informer model, an electromagnetic situation prediction framework is constructed, which can perform multi-scale analysis and time series modeling on electromagnetic signal data in a region, and realize accurate prediction of electromagnetic situation. Since the Informer model can perform real-time prediction after training, the real-time performance of electromagnetic situation prediction is improved; in addition, since incremental learning is also introduced on the basis of the prediction framework, the adaptability and flexibility of the prediction framework are further improved.

[0040] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of a flow chart of an electromagnetic situation prediction method based on an Informer model provided in an embodiment of the present invention;

[0042] Figure 2 A graph showing the error comparison results of different methods provided by an embodiment of the present invention;

[0043] Figure 3 A diagram showing a comparison of calculation times of different methods provided by an embodiment of the present invention;

[0044] Figure 4 A schematic diagram of the structure of an electromagnetic situation prediction device based on an Informer model provided in an embodiment of the present invention;

[0045] Figure 5 A schematic diagram of the structure of an electromagnetic situation prediction device based on an Informer model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0047] In order to achieve accurate and real-time prediction of electromagnetic situation, an embodiment of the present invention provides an electromagnetic situation prediction method based on an Informer model. Figure 1 A flow chart of an electromagnetic situation prediction method based on an Informer model provided in an embodiment of the present invention is shown as follows: Figure 1 As shown, including:

[0048] S101, acquiring electromagnetic signal data of a target area.

[0049] S102: performing data division processing on the electromagnetic signal data based on the time sequence to obtain a plurality of electromagnetic signal subsequences of fixed length.

[0050] Optionally, S102 may specifically include:

[0051] Sequentially dividing the electromagnetic signal data according to time to obtain a plurality of electromagnetic signal subsequences;

[0052] Normalizing the multiple electromagnetic signal subsequences to obtain multiple normalized subsequence data;

[0053] A cyclic filling method is used to perform sequence filling processing on multiple normalized subsequence data to obtain multiple electromagnetic signal subsequences with fixed lengths.

[0054] S103, inputting multiple electromagnetic signal subsequences into the current pre-trained Informer model for situation prediction processing to obtain electromagnetic situation prediction results.

[0055] It should be noted that, in the embodiment of the present invention, the current pre-trained Informer model is based on sparse attention mechanism, self-attention distillation, and generative decoder mechanism, and is obtained by incremental learning and training of electromagnetic signal sample data.

[0056] The embodiment of the present invention provides an electromagnetic situation prediction method based on the Informer model. By introducing the Informer model, an electromagnetic situation prediction framework is constructed, which can perform multi-scale analysis and time series modeling on electromagnetic signal data in a region, thereby realizing accurate prediction of the electromagnetic situation. Since the Informer model can perform real-time prediction after training, the real-time performance of the electromagnetic situation prediction is improved. In addition, since incremental learning is introduced on the basis of the prediction framework, the adaptability and flexibility of the prediction framework are further improved.

[0057] It should be noted that, in the embodiment of the present invention, a sparse attention mechanism is adopted to extract the most important attention information in the sequence, while simplifying the less important attention. This method significantly reduces the computational complexity of the model and effectively reduces the pressure on computer memory and graphics card resources. By connecting multiple layers of attention mechanisms in series to form a self-attention distillation mechanism, the attention signal that dominates the sequence processing is further highlighted, thereby providing sufficient preparation for processing longer input sequences. The introduction of a generative decoder effectively avoids the performance degradation caused by error accumulation in traditional methods, while significantly improving the computing speed.

[0058] Optionally, the current pre-trained Informer model includes: an input layer, an embedding layer, a position encoding layer, an encoder module, a self-attention distillation module, a decoder module, and an output layer connected in sequence;

[0059] The encoder module is provided with: a local attention module, a first feedforward neural network module, and a first residual connection and layer normalization module connected in sequence;

[0060] The decoder module is provided with: a self-attention module, an encoder-decoder attention module, a second feedforward neural network module, and a second residual connection and layer normalization module connected in sequence;

[0061] Among them, the local attention module is used for the setting of sparse attention mechanism, and the encoder-decoder attention module is used for the setting of generative decoder mechanism.

[0062] Optionally, the training process of the pre-trained Informer model includes:

[0063] Acquire electromagnetic signal sample data of the target area;

[0064] The electromagnetic signal sample data is divided and processed based on the time sequence to obtain a plurality of electromagnetic signal sample subsequences of fixed length;

[0065] Input multiple electromagnetic signal sample subsequences into the Informer model, and continuously train the Informer model in the direction of decreasing loss function;

[0066] The Informer model that reaches the preset stop condition is used as the pre-trained Informer model.

[0067] Optionally, the Informer model that reaches a preset stop condition is used as a pre-trained Informer model, including:

[0068] The Informer model corresponding to the number of iterations greater than the iteration threshold is used as the pre-trained Informer model; or,

[0069] The Informer model corresponding to the loss function value being less than the preset loss threshold is used as the pre-trained Informer model.

[0070] Optionally, the electromagnetic situation prediction method based on the Informer model further includes:

[0071] The data fine-tuning method is used to perform incremental learning on the pre-trained Informer model.

[0072] Optionally, a data fine-tuning method is used to perform incremental learning processing on the pre-trained Informer model, including:

[0073] Acquire new electromagnetic signal sample data;

[0074] The newly added electromagnetic signal sample data is divided and processed based on the time sequence to obtain a plurality of newly added electromagnetic signal sample subsequences of fixed length;

[0075] Freeze the parameters of the preset modules in the pre-trained Informer model to obtain a frozen Informer model;

[0076] Multiple newly added sample subsequences of electromagnetic signals are input into the frozen Informer model for incremental learning and training to obtain the current pre-trained Informer model.

[0077] Specifically, in the embodiment of the present invention, based on the trained Informer model (pre-trained Informer model), incremental learning is performed using the input new data (newly added electromagnetic signal sample data) to optimize the model and improve its adaptability in a dynamic environment. The specific implementation steps are as follows:

[0078] 1. New data preprocessing: preprocessing the newly collected electromagnetic signal sample data (data division processing in S102) to maintain data consistency.

[0079] 2. Initialize incremental learning parameters: load the parameters of the trained Informer model and freeze the parameters of some modules of the model.

[0080] Set an appropriate learning rate to balance the impact of new data on the model and maintain the original knowledge.

[0081] 3. Model update: Input the preprocessed new data into the frozen informer model, and dynamically adjust the parameter settings of the frozen informer model according to the characteristics of the new data to improve the adaptability of the model.

[0082] 4. Model performance evaluation: After obtaining the current pre-trained Informer model, evaluate the performance of the current pre-trained Informer model to ensure its prediction accuracy and generalization ability on new data.

[0083] In order to verify the effectiveness of the electromagnetic situation prediction method based on the Informer model provided in an embodiment of the present invention, a simulation experiment was also carried out. The hardware configuration and model parameter configuration are shown in Table 1 and Table 2.

[0084] Table 1 Simulation experiment hardware configuration

[0085] operating system Windows 10 21H2 CPU Intel i5-9300H GPU NVIDIA1660 Memory 8G Torch version 2.0.1 cuda version 12.2

[0086] Table 2 Simulation experiment model parameter configuration

[0087] Numeric dk 512 Initial learning rate 0.0001 Attention Count Heads 8 Number of channels 85 Epoch 10 Patients 3 Dropout 0.05

[0088] The experimental setting is to input the electromagnetic signal subsequence, and the length comparison between the start token and the predicted sequence is 1:1:1. Figure 2 This is a graph showing the error comparison results of different methods provided by the embodiment of the present invention. Figure 3 This is a graph showing the comparison of the calculation time of different methods provided by the embodiment of the present invention. Figure 2 The ordinate represents the mean absolute error (MAE). For the error comparison of different methods, the method of the present invention (informer), the Transformer method and the LSTM method are compared. For the calculation time comparison of different methods, the method of the present invention (informer) and the Transformer method are compared. Figure 2 and Figure 3 It can be seen that when predicting long sequences, the method of the present invention can achieve more reliable predictions faster and better. Compared with the Transformer model, the Informer training takes less time to achieve similar performance.

[0089] The method provided in the embodiment of the present invention can be applied to an electronic device. Specifically, the electronic device can be: a desktop computer, a portable computer, an intelligent mobile terminal, a server, etc., which is not limited in the embodiment of the present invention.

[0090] Based on the same inventive concept, an embodiment of the present invention also provides an electromagnetic situation prediction device based on the Informer model. Figure 4 The structure diagram of an electromagnetic situation prediction device based on an Informer model provided by an embodiment of the present invention is shown in FIG. Figure 4 As shown, it includes: an acquisition unit 401, a division unit 402 and a training unit 403;

[0091] The acquisition unit 401 is used to: acquire electromagnetic signal data of the target area;

[0092] The division unit 402 is used to: perform data division processing on the electromagnetic signal data based on the time sequence to obtain a plurality of electromagnetic signal subsequences of fixed length;

[0093] The training unit 403 is used to: input multiple electromagnetic signal subsequences into the current pre-trained Informer model for situation prediction processing to obtain electromagnetic situation prediction results; the current pre-trained Informer model is based on sparse attention mechanism, self-attention distillation, and generative decoder mechanism, and is obtained by incremental learning and training of electromagnetic signal sample data.

[0094] Figure 5 A structural diagram of an electromagnetic situation prediction device based on an Informer model provided in an embodiment of the present invention includes: a processor 710, a storage medium 720 and a bus 730, wherein the storage medium 720 stores machine-readable instructions executable by the processor 710, and when the electromagnetic situation prediction device based on the Informer model is running, the processor 710 communicates with the storage medium 720 through the bus 730, and the processor 710 executes the machine-readable instructions to execute the steps of the above method embodiment. The specific implementation method and technical effect are similar and will not be repeated here.

[0095] The storage medium may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk storage. Optionally, the storage medium may also be at least one storage device located away from the aforementioned processor.

[0096] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0097] It should be noted that the terms "first", "second", etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention.

[0098] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification.

[0099] Although the present invention is described herein in conjunction with various embodiments, in the process of implementing the claimed invention, those skilled in the art can understand and implement other changes of the above disclosed embodiments by viewing the drawings and the disclosed content. In the description of the present invention, the term "comprising" does not exclude other components or steps, "one" or "an" does not exclude multiple situations, and the meaning of "multiple" is two or more, unless otherwise clearly and specifically limited. In addition, certain measures are recorded in different embodiments, but this does not mean that these measures cannot be combined to produce good results.

[0100] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. An electromagnetic situation prediction method based on the Informer model, characterized in that: include: Acquire electromagnetic signal data of the target area; Performing data division processing on the electromagnetic signal data based on time sequence to obtain a plurality of electromagnetic signal subsequences of fixed length; The multiple electromagnetic signal subsequences are input into the current pre-trained Informer model for situation prediction processing to obtain an electromagnetic situation prediction result; the current pre-trained Informer model is based on a sparse attention mechanism, self-attention distillation, and a generative decoder mechanism, and is trained using incremental learning of electromagnetic signal sample data.

2. The electromagnetic situation prediction method based on the Informer model according to claim 1 is characterized in that: The step of dividing the electromagnetic signal data based on the time sequence to obtain a plurality of electromagnetic signal subsequences of fixed lengths includes: Dividing the electromagnetic signal data into sequences according to time to obtain a plurality of electromagnetic signal subsequences; Normalizing the multiple electromagnetic signal subsequences to obtain multiple normalized subsequence data; A cyclic filling method is used to perform sequence filling processing on the multiple normalized subsequence data to obtain the multiple electromagnetic signal subsequences of fixed length.

3. The electromagnetic situation prediction method based on the Informer model according to claim 1 is characterized in that: The current pre-trained Informer model includes: an input layer, an embedding layer, a position encoding layer, an encoder module, a self-attention distillation module, a decoder module, and an output layer connected in sequence; The encoder module is provided with: a local attention module, a first feedforward neural network module, and a first residual connection and layer normalization module connected in sequence; The decoder module is provided with: a self-attention module, an encoder-decoder attention module, a second feedforward neural network module, and a second residual connection and layer normalization module connected in sequence; Among them, the local attention module is used for the setting of the sparse attention mechanism, and the encoder-decoder attention module is used for the setting of the generative decoder mechanism.

4. The electromagnetic situation prediction method based on the Informer model according to claim 1 is characterized in that: The training process of the pre-trained Informer model includes: Acquire electromagnetic signal sample data of the target area; Performing data division processing on the electromagnetic signal sample data based on time sequence to obtain a plurality of electromagnetic signal sample subsequences of fixed length; Inputting the plurality of electromagnetic signal sample subsequences into an Informer model, and continuously training the Informer model in a direction in which the loss function decreases; The Informer model that reaches the preset stop condition is used as the pre-trained Informer model.

5. The electromagnetic situation prediction method based on the Informer model according to claim 1 is characterized in that: The Informer model that will reach the preset stop condition, as the pre-trained Informer model, includes: The Informer model corresponding to the time when the number of iterations is greater than the iteration threshold is used as the pre-trained Informer model; or, The Informer model corresponding to the loss function value being less than the preset loss threshold is used as the pre-trained Informer model.

6. The electromagnetic situation prediction method based on the Informer model according to claim 1 is characterized in that: The electromagnetic situation prediction method based on the Informer model also includes: The pre-trained Informer model is incrementally learned using a data fine-tuning method.

7. The electromagnetic situation prediction method based on the Informer model according to claim 6 is characterized in that: The method of using the data fine-tuning method to perform incremental learning processing on the pre-trained Informer model includes: Acquire new electromagnetic signal sample data; Performing data division processing on the newly added electromagnetic signal sample data based on time sequence to obtain a plurality of newly added electromagnetic signal sample subsequences of fixed length; Freeze parameters of the preset modules in the pre-trained Informer model to obtain a frozen Informer model; The multiple newly added sample subsequences of the electromagnetic signals are input into the frozen Informer model for incremental learning and training to obtain a current pre-trained Informer model.

8. An electromagnetic situation prediction device based on the Informer model, characterized in that: The electromagnetic situation prediction device based on the Informer model includes: an acquisition unit, a division unit and a training unit; The acquisition unit is used to: acquire electromagnetic signal data of the target area; The division unit is used to: perform data division processing on the electromagnetic signal data based on the time sequence to obtain a plurality of electromagnetic signal subsequences of fixed length; The training unit is used to: input the multiple electromagnetic signal subsequences into the current pre-trained Informer model for situation prediction processing to obtain an electromagnetic situation prediction result; the current pre-trained Informer model is based on a sparse attention mechanism, self-attention distillation, and a generative decoder mechanism, and is obtained by incremental learning and training using electromagnetic signal sample data.

9. An electromagnetic situation prediction device based on the Informer model, characterized in that: include: A processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor. When the electromagnetic situation prediction device based on the Informer model is running, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to perform the steps of the electromagnetic situation prediction method based on the Informer model as described in any one of claims 1 to 7.

Images