Method and device for determining operation strategy of power equipment

Abstract

The invention discloses a method and a device for determining an operation strategy of power equipment. The method comprises the steps that historical associated data and current temperature data of the power equipment in a preset time period are acquired, and the historical associated data comprise historical environment temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment; the historical associated data are input into a pre-trained equipment temperature prediction model, target temperature prediction data corresponding to the power equipment are determined based on a model output result of the equipment temperature prediction model, and the equipment temperature prediction model is a pre-trained time sequence model; and determining a target operation strategy corresponding to the power equipment based on the target temperature prediction data and the current temperature data. The heating condition of the power equipment is accurately predicted, and the operation strategy of the power equipment can be accurately determined in combination with the target temperature prediction data and the current temperature data.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment state detection and prediction, and in particular to a method and device for determining an operation strategy of power equipment. Background Art

[0002] With the continuous development of the power system, the operation safety of power equipment has become the key to ensuring power supply. During the operation of power equipment, the resistance heat, electromagnetic induction, eddy current and other factors generated by the current passing through the conductor will cause the equipment to heat up. Long-term high-temperature operation will not only reduce the service life of the equipment, but may also cause equipment failures and even fires and other safety accidents. Therefore, it is of great significance to predict the heating of power equipment and discover potential safety hazards in time to ensure the safe and stable operation of the power system.

[0003] Traditional methods for predicting heating of power equipment mostly rely on temperature difference threshold judgment, but this method cannot effectively track and predict the future heating trend of the equipment. Although deep learning models such as long short-term memory networks have been applied to the prediction of hot spot temperature of substation equipment, long short-term memory networks have deficiencies in dealing with long-term dependencies, resulting in poor prediction of long time series and inability to accurately predict the temperature of power equipment, resulting in the inability to accurately determine the operation strategy of power equipment, affecting the safe operation of power equipment. Summary of the invention

[0004] The present invention provides a method and device for determining an operation strategy of an electric device, so as to solve the problem that the temperature of the electric device cannot be accurately predicted, resulting in the inability to accurately determine the operation strategy of the electric device, thus affecting the safe operation of the electric device.

[0005] According to one aspect of the present invention, a method for determining an operation strategy of an electric power device is provided, the method comprising:

[0006] Acquire historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment;

[0007] Inputting the historical correlation data into a pre-trained device temperature prediction model, and determining target temperature prediction data corresponding to the power device based on a model output result of the device temperature prediction model, wherein the device temperature prediction model is a pre-trained time series model;

[0008] A target operation strategy corresponding to the electric power equipment is determined based on the target temperature prediction data and the current temperature data.

[0009] According to another aspect of the present invention, there is provided a device for determining an operation strategy of an electric power device, the device comprising:

[0010] A data acquisition module, used to acquire historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment;

[0011] A temperature prediction module, used for inputting the historical correlation data into a pre-trained device temperature prediction model, and determining target temperature prediction data corresponding to the power device based on a model output result of the device temperature prediction model, wherein the device temperature prediction model is a pre-trained time series model;

[0012] A strategy determination module is used to determine an operation strategy corresponding to the power equipment based on the target temperature prediction data and the current temperature data.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for determining an operation strategy of an electric power device described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for determining an operation strategy of an electric power device described in any embodiment of the present invention when executed.

[0018] The technical solution of the embodiment of the present invention, by obtaining the historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment; it is convenient to accurately predict the target temperature prediction data of the equipment based on the historical associated data. Then, the historical associated data is input into the pre-trained equipment temperature prediction model, and the target temperature prediction data corresponding to the power equipment is determined based on the model output result of the equipment temperature prediction model, wherein the equipment temperature prediction model is a pre-trained time series model; the target temperature prediction data of the equipment is accurately predicted, and finally, the target operation strategy corresponding to the power equipment is determined based on the target temperature prediction data and the current temperature data, which solves the problem that the temperature of the power equipment cannot be accurately predicted, resulting in the inability to accurately determine the operation strategy of the power equipment, affecting the safe operation of the power equipment, and realizes accurate prediction of the heating situation of the power equipment. The operation strategy of the power equipment can be accurately determined by combining the target temperature prediction data and the current temperature data. The accuracy of the temperature prediction data of the power equipment is improved, the accuracy of determining the operation strategy of the power equipment is improved, and the beneficial effect of ensuring the safe operation of the power equipment is achieved.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flow chart of a method for determining an operation strategy of an electric power device provided according to a first embodiment of the present invention;

[0022] Figure 2 is a flow chart of a method for determining an operation strategy of an electric power device provided in accordance with a second embodiment of the present invention;

[0023] Figure 3 is a structural diagram of a device for determining an operation strategy of an electric power device provided according to a third embodiment of the present invention;

[0024] Figure 4 It is a structural schematic diagram of an electronic device for implementing the method for determining an operation strategy of an electric power device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings 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 interchanged 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. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] Embodiment 1

[0028] Figure 1 A flowchart of a method for determining an operation strategy of an electric device is provided for the first embodiment of the present invention. This embodiment is applicable to the case of determining an operation strategy of an electric device. The method can be executed by an operation strategy determining device of an electric device. The operation strategy determining device of the electric device can be implemented in the form of hardware and / or software. The operation strategy determining device of the electric device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0029] S110. Acquire historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data, and historical temperature data of the power equipment.

[0030] The historical correlation data may be understood as historical temperature correlation data. The current temperature data may be understood as current actual temperature data. The preset time period may be preset based on experience, such as 96 hours, which is not limited in this embodiment.

[0031] Specifically, historical related data of power equipment is collected, including historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of power equipment, etc., and the historical related data is preprocessed, and the preprocessing includes standardization processing to ensure the quality and consistency of the historical related data.

[0032] S120. Input the historical correlation data into a pre-trained equipment temperature prediction model, and determine the target temperature prediction data corresponding to the power equipment based on the model output result of the equipment temperature prediction model, wherein the equipment temperature prediction model is a pre-trained time series model.

[0033] The target temperature prediction data may be understood as temperature prediction data of the power equipment, and the target temperature prediction data includes a target temperature prediction data curve constructed by temperature prediction data at at least one time point.

[0034] Specifically, an improved time series model network architecture is designed, and the dilated causal convolution operation is used in the distillation operation. At the same time, residual connections are introduced between attention blocks to capture the key feature information in the time series and make predictions; the constructed improved time series model uses the kernel mean square error function to calculate the loss to optimize the nonlinear problem of model loss, enhance the network's learning ability for complex time series data, and improve prediction accuracy and stability; collect a preset number of historical correlation data to construct a data set, divide the data set into a training set and a test set, and train the constructed model based on the training set to obtain the optimal values ​​of the model parameters of the equipment temperature prediction model; the trained equipment temperature prediction model is evaluated using the test set, and the prediction performance is quantified by calculating indicators such as mean square error. Verify the effectiveness of the introduction of extended causal convolution, residual connection, and kernel mean square error function in improving prediction performance. The equipment temperature prediction model is deployed in the actual power equipment monitoring system to predict the equipment temperature.

[0035] Exemplarily, 96 hours of historical correlation data within the historical time period are input into the equipment temperature prediction model, and the temperature prediction data for the next 24 consecutive hours are predicted based on the equipment temperature prediction model. The 24 hours of temperature prediction data are arranged in chronological order and a temperature prediction curve is generated.

[0036] Optionally, the device temperature prediction model includes an encoder and a decoder connected to the encoder; the encoder includes an input layer, an embedding layer, a residual module and an output layer connected in series in sequence, wherein the residual module includes multiple residual units, at least two residual units are residually connected, each residual unit includes a self-attention block and an expanded causal convolution layer connected to the self-attention block, and the output layer is respectively connected to the self-attention block of the first residual unit and the expanded causal convolution layer of the last residual unit; the decoder includes a multi-head self-attention layer and a fully connected layer.

[0037] Specifically, the encoder uses a multi-head probabilistic sparse self-attention mechanism to efficiently identify key dependencies in the input historical data sequence while reducing computational complexity. Subsequently, a refined self-attention feature map is constructed through a distillation operation to further refine the target operation feature data. The decoder processes the target operation feature data containing position and time features through a similar mechanism and combines it with the encoder output to ensure the causality of the prediction process. Finally, the fully connected layer adjusts the output dimension to generate high-precision target temperature prediction data.

[0038] Optionally, the historical correlation data is input into a pre-trained device temperature prediction model, and the temperature prediction data is determined based on the model output result, including: determining a historical data sequence corresponding to the historical correlation data, inputting the historical data sequence into the encoder to obtain target operating characteristic data corresponding to the historical correlation data output by the encoder; and inputting the target operating characteristic data into the decoder to obtain target temperature prediction data corresponding to the target operating characteristic data output by the decoder.

[0039] The historical data sequence can be understood as a sequence obtained by arranging the historical data in chronological order. The target operation characteristic data can be understood as key operation characteristic data.

[0040] Specifically, the historical ambient temperature data, historical wind speed data, historical load current data, and historical temperature data of the power equipment within a preset time period are arranged in chronological order to form a historical data sequence. The historical data sequence is input into an encoder. The encoder encodes the input historical data sequence and extracts the target operation characteristic data therein. The target operation characteristic data output by the encoder is input into a decoder, the target operation characteristic data extracted by the encoder is decoded, and target temperature prediction data corresponding to the target operation characteristic data is generated.

[0041] In the embodiment of the present invention, determining the target temperature prediction data through the trained equipment temperature prediction model not only reduces the computational burden of the model in processing long time series, but also significantly improves the accuracy of the prediction, and is an effective tool for processing complex time series prediction tasks.

[0042] Optionally, inputting the historical data sequence into the encoder to obtain target operation feature data corresponding to the historical associated data output by the encoder includes: inputting the historical data sequence into the input layer, converting the data sequence into an embedding vector through the embedding layer connected to the input layer; and inputting the embedding vector into a residual module to obtain the target operation feature data.

[0043] Specifically, the historical data sequence is input into the input layer of the model, and the data of the input layer is passed to the embedding layer to convert the input data sequence into an embedding vector. In the residual module, the embedding vector is subjected to multi-layer convolution, activation function, residual connection and other operations to gradually extract the target operation feature data in the data. The output layer receives the target operation feature data output by the residual module, processes the target operation feature data (such as linear transformation, activation function, etc.), and outputs it as the final prediction result.

[0044] Optionally, inputting the embedding vector into the residual module to obtain target running feature data includes: inputting the embedding vector into the residual module, determining the target output vector corresponding to the embedding vector for each residual unit through the self-attention block, and extracting initial running feature data in the target output vector through the dilated causal convolution layer connected to the self-attention block; transmitting the original output vector output by the self-attention block of the first residual unit and the initial running feature data output by the dilated causal convolution layer of the last residual unit as target feature data to the output layer, wherein the target output vector includes a query vector, a key vector and a value vector.

[0045] Specifically, for each residual unit, a self-attention block is used to process the embedding vector, which converts the embedding vector into a query vector, a key vector, and a value vector. The vectors interact through a dot product attention mechanism to calculate the attention weights for each position and generate a target output vector. The target output vector output by the self-attention block is input into the dilated causal convolution layer. By increasing the spacing of the convolution kernel, the dilated convolution can capture long-range dependencies while maintaining computational efficiency. Causal convolution ensures that only information at the current position and before can be used when processing sequence data to avoid information leakage. The dilated causal convolution layer extracts initial running feature data from the target output vector. The original output vector output by the self-attention block of the first residual unit and the initial running feature data output by the dilated causal convolution layer of the last residual unit are fused to form the target feature data. The target feature data contains the deep features and global dependencies of the input sequence.

[0046] In an embodiment of the present invention, in the encoder, dilated causal convolution is used to replace regular convolution to increase the receptive field of the model; at the same time, residual connection is introduced to enhance the stability and robustness of the model.

[0047] Optionally, firstly, a multi-head probabilistic sparse self-attention mechanism is used to capture the key dependencies in the input sequence while reducing the computational cost. Secondly, a focused self-attention feature map is constructed through a distillation operation to extract key information. The output vector corresponding to the embedding vector determined by the multi-head self-attention layer is expressed as follows:

[0048]

[0049] Among them, A(Q, K, V) represents the target output vector, Q represents the query vector after probability sparsification, K represents the key vector, and V represents the value vector.

[0050] The distillation operation is performed according to the following formula:

[0051]

[0052] in,[ ] AB Represents a self-attention block, Convld() represents a one-dimensional convolution, and ReLU() represents an activation function.

[0053] The activation function performs a regular convolution operation in the time dimension, and adds a maximum pooling layer after each convolution layer to downsample it to half, thereby reducing the memory usage of the entire model.

[0054] Optionally, the initial running feature data in the target output vector extracted by the dilated causal convolutional layer connected to the self-attention block is expressed by the following formula:

[0055]

[0056] Among them, C(W, x t ) represents the initial operation characteristic data, w k represents a convolution kernel of size K, x represents the historical correlation data at time t in the input historical data sequence, and d represents the expansion coefficient.

[0057] Optionally, the target temperature prediction data corresponding to the target operation characteristic data is determined by the decoder, including: generating initial temperature prediction data corresponding to the target operation characteristic data through the multi-head self-attention layer; adjusting the dimension of the initial temperature prediction data through the fully connected layer to obtain the target temperature prediction data corresponding to the target operation characteristic data.

[0058] Specifically, the decoder masks part of the input data with position features and time features and sends it to the multi-head probabilistic sparse self-attention operation, and then performs a multi-head self-attention operation with the intermediate result output by the encoder. Finally, the data output dimension is adjusted through the fully connected layer to obtain the target temperature prediction data.

[0059] S130. Determine a target operation strategy corresponding to the power equipment based on the target temperature prediction data and the current temperature data.

[0060] The target operation strategy can be understood as the operation strategy to be executed by the power equipment.

[0061] Specifically, the trend and magnitude of future temperature changes can be evaluated by calculating the difference between the target temperature prediction data and the current temperature data. Based on the target temperature prediction data, the temperature change trend of the power equipment in the future period is analyzed, including rising, falling or remaining stable. If it is predicted that the temperature will rise significantly in the future, it may be necessary to adjust the operating parameters of the power equipment, such as reducing the load, increasing heat dissipation, etc., to prevent the equipment from overheating. If it is predicted that the temperature will drop in the future, the operating status of the equipment can be adjusted accordingly to improve efficiency or save energy.

[0062] The technical solution of the embodiment of the present invention, by obtaining the historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment; it is convenient to accurately predict the target temperature prediction data of the equipment based on the historical associated data. Then, the historical associated data is input into the pre-trained equipment temperature prediction model, and the target temperature prediction data corresponding to the power equipment is determined based on the model output result of the equipment temperature prediction model, wherein the equipment temperature prediction model is a pre-trained time series model; the target temperature prediction data of the equipment is accurately predicted, and finally, the target operation strategy corresponding to the power equipment is determined based on the target temperature prediction data and the current temperature data, which solves the problem that the temperature of the power equipment cannot be accurately predicted, resulting in the inability to accurately determine the operation strategy of the power equipment, affecting the safe operation of the power equipment, and realizes accurate prediction of the heating situation of the power equipment. The operation strategy of the power equipment can be accurately determined by combining the target temperature prediction data and the current temperature data. The accuracy of the temperature prediction data of the power equipment is improved, the accuracy of determining the operation strategy of the power equipment is improved, and the beneficial effect of ensuring the safe operation of the power equipment is achieved.

[0063] Embodiment 2

[0064] Figure 2A flowchart of a method for determining an operation strategy of an electric power device provided in Embodiment 2 of the present invention, this embodiment is a further refinement of how to determine the operation strategy corresponding to the electric power device based on the target temperature prediction data and the current temperature data in the above embodiment. Optionally, the operation strategy corresponding to the electric power device based on the target temperature prediction data and the current temperature data includes: comparing the target temperature prediction data with the current operating temperature data, and determining the target operation strategy corresponding to the electric power device based on the comparison result, wherein the target operation strategy includes a preventive maintenance strategy, a load management strategy, and a cooling adjustment strategy.

[0065] like Figure 2 As shown, the method includes:

[0066] S210. Acquire historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data, and historical temperature data of the power equipment.

[0067] S220. Input the historical correlation data into a pre-trained equipment temperature prediction model, and determine the target temperature prediction data corresponding to the power equipment based on the model output result of the equipment temperature prediction model, wherein the equipment temperature prediction model is a pre-trained time series model.

[0068] S230. Compare the target temperature prediction data with the current operating temperature data, and determine a target operating strategy corresponding to the power equipment based on the comparison result, wherein the target operating strategy includes a preventive maintenance strategy, a load management strategy, and a cooling adjustment strategy.

[0069] Specifically, the target temperature prediction data is compared with the current operating temperature data in time series to analyze the differences and change trends between the two. If it is predicted that the temperature will rise significantly in the future and may exceed the normal operating range of the equipment, then preventive maintenance needs to be carried out in advance, such as checking the equipment's cooling system, cleaning dust, replacing aging components, etc. According to the trend of the temperature prediction data, a more accurate maintenance plan can be formulated to avoid emergency maintenance when the equipment temperature reaches the critical value. If it is predicted that the temperature will rise in the future due to excessive load, then the load of the equipment needs to be adjusted, such as reducing the load, redistributing the load, or optimizing the load mode. Through load management, the temperature and performance of the equipment can be balanced to avoid performance degradation or failure caused by overheating of the equipment. If it is predicted that the temperature will rise in the future and the current cooling system is not sufficient to maintain the equipment within the normal operating range, the cooling strategy needs to be adjusted, such as increasing the speed of the cooling fan, adjusting the flow rate or temperature of the cooling water, etc. Through cooling adjustment, the equipment temperature can be effectively reduced and the stability and reliability of the equipment can be improved. According to the formulated target operation strategy, the operating parameters, maintenance plan or cooling system of the power equipment are adjusted. After the strategy is implemented for a period of time, its effect on equipment temperature control, energy efficiency optimization and reliability improvement is evaluated. Based on the evaluation results and new data input, the target operation strategy is continuously improved and optimized to adapt to changes in the equipment operating environment and changes in performance requirements.

[0070] The technical solution of the embodiment of the present invention compares the target temperature prediction data with the current operating temperature data, and determines the target operating strategy corresponding to the power equipment based on the comparison result, wherein the target operating strategy includes a preventive maintenance strategy, a load management strategy, and a cooling adjustment strategy. Based on the comparison result of the target temperature prediction data and the current operating temperature data, a more accurate and effective target operating strategy is formulated for the power equipment to improve the reliability, energy efficiency, and safety of the equipment.

[0071] As an optional example of an embodiment of the present invention, the method for determining an operation strategy of an electric power device in this embodiment specifically includes the following steps:

[0072] Step 1) obtaining historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment, and performing standardization processing on the data to ensure the quality and consistency of the data;

[0073] Step 2) Construct a time series model. Based on the improved time series algorithm, a time series model is constructed. The model adopts an encoder-decoder structure, where the encoder is used to extract the features of the input sequence and the decoder is used to realize sequence prediction. In the encoder, the dilated causal convolution is used to replace the regular convolution to increase the receptive field of the model; at the same time, the residual connection is introduced to enhance the stability and robustness of the model. In terms of the loss function, the kernel mean square error function is used to deal with the nonlinear problem of the error.

[0074] Specifically, an improved time series model network architecture is designed, and the dilated causal convolution operation is used in the distillation operation. At the same time, residual connections are introduced between attention blocks to capture the key feature information in the time series for prediction. The constructed improved time series model uses the kernel mean square error function to calculate the loss to optimize the nonlinear problem of model loss, enhance the network's learning ability for complex time series data, and improve prediction accuracy and stability.

[0075] Step 3) Collect a preset amount of historical correlation data to construct a data set, divide the data set into a training set and a test set, and train the constructed model based on the training set to obtain the optimal values ​​of the model parameters of the equipment temperature prediction model.

[0076] Step 4) The trained device temperature prediction model is evaluated using the test set, and the prediction performance is quantified by calculating indicators such as mean square error. This verifies the effectiveness of the extended causal convolution, residual connection, and kernel mean square error function in improving the prediction performance.

[0077] Step 5) Deploy the trained equipment temperature prediction model to the actual power equipment monitoring system to predict the equipment temperature. Combine the prediction results with the actual monitoring data to adjust the operation strategy of the power equipment in real time to prevent overheating and improve the safety and reliability of equipment operation. Deploy the equipment temperature prediction model to the actual power equipment monitoring system to predict the equipment temperature.

[0078] The equipment temperature prediction model is based on the improved time series algorithm, and is specifically established as follows:

[0079] The time series algorithm adopts an encoder and decoder structure design, in which the encoder adopts a multi-head probabilistic sparse self-attention mechanism and a distillation operation, while the decoder introduces a multi-head probabilistic sparse self-attention mechanism with a mask. The encoder includes an input layer, an embedding layer, a residual module, and an output layer connected in series, wherein the residual module includes a plurality of residual units, at least two residual units are residually connected, each residual unit includes a self-attention block and an extended causal convolution layer connected to the self-attention block, and the output layer is respectively connected to the self-attention block of the first residual unit and the extended causal convolution layer of the last residual unit; the decoder includes a multi-head self-attention layer and a fully connected layer.

[0080] Encoder: First, the key dependencies in the input sequence are captured through a multi-head probabilistic sparse self-attention mechanism while reducing the computational cost. Secondly, a focused self-attention feature map is constructed through a distillation operation to extract key information. By inputting the embedding vector into the residual module, for each residual unit, the target output vector corresponding to the embedding vector is determined through the self-attention block, and the initial running feature data in the target output vector is extracted through the dilated causal convolution layer connected to the self-attention block; the original output vector output by the self-attention block of the first residual unit and the initial running feature data output by the dilated causal convolution layer of the last residual unit are transmitted to the output layer as target feature data, wherein the target output vector includes a query vector, a key vector, and a value vector.

[0081] First, the key dependencies in the input sequence are captured through the multi-head probabilistic sparse self-attention mechanism, while reducing the computational cost. Secondly, a focused self-attention feature map is constructed through a distillation operation to extract key information. The output vector corresponding to the embedding vector determined by the multi-head self-attention layer is expressed as follows:

[0082]

[0083] Among them, A(Q, K, V) represents the target output vector, Q represents the query vector after probability sparsification, K represents the key vector, and V represents the value vector.

[0084] The distillation operation is performed according to the following formula:

[0085]

[0086] in,[ ] AB Represents a self-attention block, Convld() represents a one-dimensional convolution, and ReLU() represents an activation function.

[0087] The decoder will mask the feature data of some targets with position features and time features and send them into the multi-head probabilistic sparse self-attention operation, and then perform the multi-head self-attention operation with the intermediate result output by the encoder. Finally, the data output dimension is adjusted through the fully connected layer to obtain the prediction result.

[0088] The encoder efficiently captures the key dependencies in the input sequence and extracts key information through the multi-head probabilistic sparse self-attention mechanism and distillation operation. The decoder uses the masked multi-head probabilistic sparse self-attention mechanism to ensure the causality of the prediction, and generates the final prediction result by combining it with the encoder output and adjusting the fully connected layer. When dealing with long time series prediction problems, it can effectively reduce the computational cost and improve the prediction accuracy.

[0089] The equipment temperature prediction model is as follows:

[0090] Based on the time series model, the distillation operation layer is modified to optimize the convolution operation. The specific operation is to replace the original regular convolution with the dilated causal convolution to expand the receptive field of the model, so that the model can extract features from longer historical information. Secondly, residual connections are added to the stacking layer to reduce the possibility of network degradation. Finally, the kernel mean square error function is used to deal with the nonlinear problem of error.

[0091] The operation feature data in the output vector is extracted by the dilated causal convolution layer, which is expressed by the following formula:

[0092]

[0093] Among them, C(W, x t ) represents the initial operation characteristic data, w k represents a convolution kernel of size K, x represents the historical correlation data at time t in the input historical data sequence, and d represents the expansion coefficient.

[0094] Residual connection: In the device temperature prediction model, the encoder's self-attention blocks and convolutional pooling layers are stacked on each other to increase the network depth and enhance the robustness of the model. However, the stacking of self-attention blocks and convolutional pooling layers may cause network degradation problems. As the depth of the network increases, the original information obtained by the network will decrease layer by layer, which will affect the stability of the model training. In severe cases, gradient disappearance or explosion problems may occur. In order to weaken this effect, the technical solution of an embodiment of the present invention adds a residual connection to the encoder to map the information output by the first self-attention block directly to the last layer, ensuring that even if the network depth is stacked, the output layer can still receive a certain amount of original information from the input layer, so as to increase the diversity of model extraction features while enhancing the stability of model training.

[0095] Kernel mean square error function: The technical solution of the embodiment of the present invention introduces a nonlinear mapping function based on the pre-trained time series model, which maps the error from the original feature space to the higher-dimensional reproducing kernel Hilbert space, so as to accurately quantify the prediction error in the nonlinear domain. The kernel function technology, especially the Gaussian kernel function, is used because it is a radial extension of the Euclidean measure and can keep the relative position between data unchanged, thus becoming the choice of this method.

[0096] The technical solution of the embodiment of the present invention not only inherits the advantages of the time series model in long time series prediction, but also introduces dilated causal convolution, residual connection and uses kernel mean square error function, which further improves the accuracy and robustness of the model in processing power equipment heating prediction. It solves the problem that the temperature of power equipment cannot be accurately predicted, resulting in the inability to accurately determine the operating strategy of the power equipment, affecting the safe operation of the power equipment, and achieves accurate prediction of the heating situation of the power equipment. The operating strategy of the power equipment can be accurately determined by combining the target temperature prediction data and the current temperature data. It achieves the beneficial effect of improving the accuracy of the power equipment temperature prediction data, improving the accuracy of determining the power equipment operating strategy, and ensuring the safe operation of the power equipment.

[0097] Embodiment 3

[0098] Figure 3 This is a schematic diagram of the structure of a device for determining an operation strategy of an electric power device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: a data acquisition module 310 , a temperature prediction module 320 and a strategy determination module 330 .

[0099] Among them, the data acquisition module 310 is used to obtain the historical related data and current temperature data of the power equipment within a preset time period, wherein the historical related data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment; the temperature prediction module 320 is used to input the historical related data into a pre-trained equipment temperature prediction model, and determine the target temperature prediction data corresponding to the power equipment based on the model output result of the equipment temperature prediction model, wherein the equipment temperature prediction model is a pre-trained time series model; the strategy determination module 330 is used to determine the operation strategy corresponding to the power equipment based on the target temperature prediction data and the current temperature data.

[0100] The technical solution of the embodiment of the present invention obtains the historical associated data and current temperature data of the power equipment within a preset time period through a data acquisition module, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment; it is convenient to accurately predict the target temperature prediction data of the equipment based on the historical associated data in the subsequent period. Then, through the temperature prediction module, the historical associated data is input into the pre-trained equipment temperature prediction model, and the target temperature prediction data corresponding to the power equipment is determined based on the model output result of the equipment temperature prediction model, wherein the equipment temperature prediction model is a pre-trained time series model; the target temperature prediction data of the equipment is accurately predicted, and finally, through the strategy determination module, the target operation strategy corresponding to the power equipment is determined based on the target temperature prediction data and the current temperature data, so as to solve the problem that the temperature of the power equipment cannot be accurately predicted, resulting in the inability to accurately determine the operation strategy of the power equipment, affecting the safe operation of the power equipment, and realizing the accurate prediction of the heating situation of the power equipment. The operation strategy of the power equipment can be accurately determined by combining the target temperature prediction data and the current temperature data. The accuracy of the temperature prediction data of the power equipment is improved, the accuracy of determining the operation strategy of the power equipment is improved, and the beneficial effect of ensuring the safe operation of the power equipment is achieved.

[0101] Optionally, the temperature prediction module is specifically used for:

[0102] The equipment temperature prediction model includes an encoder and a decoder connected to the encoder; the encoder includes an input layer, an embedding layer, a residual module and an output layer connected in series in sequence, wherein the residual module includes multiple residual units, at least two residual units are residually connected, each residual unit includes a self-attention block and an expanded causal convolution layer connected to the self-attention block, and the output layer is respectively connected to the self-attention block of the first residual unit and the expanded causal convolution layer of the last residual unit; the decoder includes a multi-head self-attention layer and a fully connected layer.

[0103] Optionally, the temperature prediction module includes:

[0104] a target operation characteristic acquisition unit, used for determining a historical data sequence corresponding to the historical association data, and inputting the historical data sequence into the encoder to obtain target operation characteristic data corresponding to the historical association data output by the encoder;

[0105] The target temperature prediction data acquisition unit is used to input the target operation characteristic data into the decoder to obtain the target temperature prediction data corresponding to the target operation characteristic data output by the decoder.

[0106] Optionally, the target operation feature acquisition unit includes:

[0107] a vector conversion subunit, configured to input the historical data sequence into the input layer, and convert the data sequence into an embedded vector through the embedding layer connected to the input layer;

[0108] The target operation feature acquisition subunit is used to input the embedding vector into the residual module to obtain the target operation feature data.

[0109] Optionally, the target operation feature acquisition subunit is specifically used for:

[0110] By inputting the embedding vector into the residual module, for each residual unit, determining a target output vector corresponding to the embedding vector through the self-attention block, and extracting initial running feature data in the target output vector through the dilated causal convolution layer connected to the self-attention block;

[0111] The original output vector output by the self-attention block of the first residual unit and the initial running feature data output by the dilated causal convolutional layer of the last residual unit are transmitted to the output layer as target feature data, wherein the target output vector includes a query vector, a key vector and a value vector.

[0112] Optionally, the temperature prediction module is specifically used for:

[0113] The output vector corresponding to the embedding vector determined by the multi-head self-attention layer is expressed by the following formula:

[0114]

[0115] Among them, A(Q, K, V) represents the target output vector, Q represents the query vector after probability sparsification, K represents the key vector, and V represents the value vector.

[0116] Optionally, the target operation feature acquisition subunit is specifically used for:

[0117] The initial running feature data in the target output vector is extracted by the dilated causal convolutional layer connected to the self-attention block, which is expressed as follows:

[0118]

[0119] Among them, C(W, x t ) represents the initial operation characteristic data, w k represents a convolution kernel of size K, x represents the historical correlation data at time t in the input historical data sequence, and d represents the expansion coefficient.

[0120] Optionally, the temperature prediction module includes:

[0121] An initial temperature prediction data acquisition unit, configured to generate initial temperature prediction data corresponding to the target operation characteristic data through the multi-head self-attention layer;

[0122] a target temperature prediction data acquisition unit, configured to adjust the dimension of the initial temperature prediction data through the fully connected layer to obtain target temperature prediction data corresponding to the target operation characteristic data. Optionally, the strategy determination module includes:

[0123] The target temperature prediction data and the current operating temperature data are compared, and a target operating strategy corresponding to the power equipment is determined based on the comparison result, wherein the target operating strategy includes a preventive maintenance strategy, a load management strategy and a cooling adjustment strategy.

[0124] The device for determining the operation strategy of electric power equipment provided in the embodiment of the present invention can execute the method for determining the operation strategy of electric power equipment provided in any embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method.

[0125] Embodiment 4

[0126] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0127] like Figure 4As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0128] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0129] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as the method for determining the operation strategy of the power equipment.

[0130] In some embodiments, the method for determining the operation strategy of the power device can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for determining the operation strategy of the power device described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the method for determining the operation strategy of the power device in any other appropriate manner (for example, by means of firmware).

[0131] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0132] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0133] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0134] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0135] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0136] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0137] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0138] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for determining an operation strategy of an electric power device, characterized in that: include: Acquire historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment; Inputting the historical correlation data into a pre-trained device temperature prediction model, and determining target temperature prediction data corresponding to the power device based on a model output result of the device temperature prediction model, wherein the device temperature prediction model is a pre-trained time series model; A target operation strategy corresponding to the electric power equipment is determined based on the target temperature prediction data and the current temperature data.

2. The method according to claim 1, characterized in that The equipment temperature prediction model includes an encoder and a decoder connected to the encoder; the encoder includes an input layer, an embedding layer, a residual module and an output layer connected in series in sequence, wherein the residual module includes multiple residual units, at least two residual units are residually connected, each residual unit includes a self-attention block and an expanded causal convolution layer connected to the self-attention block, and the output layer is respectively connected to the self-attention block of the first residual unit and the expanded causal convolution layer of the last residual unit; the decoder includes a multi-head self-attention layer and a fully connected layer.

3. The method according to claim 2, characterized in that The step of inputting the historical correlation data into a pre-trained device temperature prediction model and determining temperature prediction data based on the model output result includes: Determine a historical data sequence corresponding to the historical association data, and input the historical data sequence into the encoder to obtain target operation characteristic data corresponding to the historical association data output by the encoder; The target operation characteristic data is input into the decoder to obtain target temperature prediction data corresponding to the target operation characteristic data output by the decoder.

4. The method according to claim 2, characterized in that: The step of inputting the historical data sequence into the encoder to obtain target operation characteristic data output by the encoder corresponding to the historical associated data includes: Inputting the historical data sequence into the input layer, and converting the data sequence into an embedding vector through the embedding layer connected to the input layer; The embedding vector is input into the residual module to obtain the target operation feature data.

5. The method according to claim 2, characterized in that: The step of inputting the embedding vector into a residual module to obtain target operation feature data includes: By inputting the embedding vector into the residual module, for each residual unit, determining a target output vector corresponding to the embedding vector through the self-attention block, and extracting initial running feature data in the target output vector through the dilated causal convolution layer connected to the self-attention block; The original output vector output by the self-attention block of the first residual unit and the initial running feature data output by the dilated causal convolutional layer of the last residual unit are transmitted to the output layer as target feature data, wherein the target output vector includes a query vector, a key vector and a value vector.

6. The method according to claim 4, characterized in that The target output vector corresponding to the embedding vector determined by the self-attention block is expressed by the following formula: Among them, A(Q, K, V) represents the target output vector, Q represents the query vector after probability sparsification, K represents the key vector, and V represents the value vector.

7. The method according to claim 4, characterized in that The initial running feature data extracted from the target output vector by the dilated causal convolutional layer connected to the self-attention block is expressed as follows: Among them, C(W, x t ) represents the initial operation characteristic data, w k represents a convolution kernel of size K, x represents the historical correlation data at time t in the input historical data sequence, and d represents the expansion coefficient.

8. The method according to claim 2, characterized in that: Determining target temperature prediction data corresponding to the target operation characteristic data by the decoder includes: Generate initial temperature prediction data corresponding to the target operation feature data through the multi-head self-attention layer; The dimension of the initial temperature prediction data is adjusted through the fully connected layer to obtain target temperature prediction data corresponding to the target operation characteristic data.

9. The method according to claim 1, characterized in that: The determining an operation strategy corresponding to the electric power equipment based on the target temperature prediction data and the current temperature data comprises: The target temperature prediction data and the current operating temperature data are compared, and a target operating strategy corresponding to the power equipment is determined based on the comparison result, wherein the target operating strategy includes a preventive maintenance strategy, a load management strategy and a cooling adjustment strategy.

10. A device for determining an operation strategy of an electric power device, characterized in that: include: A data acquisition module, used to acquire historical associated data and current temperature data of the power equipment within a preset time period, wherein the historical associated data includes historical ambient temperature data, historical wind speed data, historical load current data and historical temperature data of the power equipment; A temperature prediction module, used for inputting the historical correlation data into a pre-trained device temperature prediction model, and determining target temperature prediction data corresponding to the power device based on a model output result of the device temperature prediction model, wherein the device temperature prediction model is a pre-trained time series model; A strategy determination module is used to determine an operation strategy corresponding to the power equipment based on the target temperature prediction data and the current temperature data.

Images