Power system scheduling method and device based on risk assessment, electronic equipment and storage medium

Abstract

The invention discloses an electric power system scheduling method and device based on risk assessment, electronic equipment and a storage medium. The method comprises the steps of obtaining scheduling risk related data of an electric power scheduling system; and inputting the scheduling risk related data to a preset power scheduling system risk assessment model, so that the power scheduling system risk assessment model performs feature extraction on the input scheduling risk related data, and then assesses the power scheduling risk of the power scheduling system according to the extracted features. Outputting a power dispatching risk score corresponding to the power dispatching system; and optimizing a current power dispatching strategy according to the power dispatching risk score, and then carrying out power system dispatching according to the optimized power dispatching strategy. According to the invention, the efficiency and accuracy of power dispatching can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power dispatching, and in particular to a method, device, electronic equipment and storage medium for power system dispatching based on risk assessment. Background Art

[0002] Power system dispatch is the core link to ensure the safe, stable and economical operation of the power grid. With the continuous expansion of the scale of the power system and the widespread access to new energy sources, the complexity and uncertainty of power grid operation have increased significantly. Traditional power system dispatch mainly relies on the experience of dispatchers to deal with the operation and control problems of the power grid. Although these methods have historically provided a strong guarantee for the stable operation of the power system, they have gradually shown some limitations in the current rapidly developing technological context. First, the dispatching decision of the existing power dispatching system depends on the personal experience of the dispatcher, and the dispatcher may not be able to make the best response in time in an emergency. Secondly, the existing system does not adequately assess the risks of power dispatching when conducting power dispatching, and is unable to identify and quantify the potential risks in the operation of the power grid, which limits the system's ability to prevent and manage risks, and may lead to a slow response when the system faces complex power grid failures or anomalies. Summary of the invention

[0003] The present invention provides a power system dispatching method, device, electronic device and storage medium based on risk assessment to solve the technical problems that the dispatching decision of the existing power dispatching system depends on the personal experience of the dispatcher and the assessment of power dispatching risks is insufficient when performing power dispatching.

[0004] In order to solve the above technical problems, an embodiment of the present invention provides a power system dispatching method based on risk assessment, comprising:

[0005] Obtaining dispatch risk related data of the power dispatch system; wherein the dispatch risk related data includes: operation ticket data, personnel duty data and equipment data;

[0006] Inputting the dispatch risk related data into a preset power dispatch system risk assessment model, so that the power dispatch system risk assessment model extracts features from the input dispatch risk related data, and then assesses the power dispatch risk of the power dispatch system according to the extracted features, and outputs a power dispatch risk score corresponding to the power dispatch system;

[0007] Optimizing the current power dispatching strategy according to the power dispatching risk score, and then dispatching the power system according to the optimized power dispatching strategy;

[0008] Among them, the power dispatching system risk assessment model is obtained by training a preset ResNet neural network model with historical dispatching risk related data of the power dispatching system as input and the power dispatching risk score corresponding to the historical dispatching risk related data as output.

[0009] As a preferred solution, the risk assessment model of the power dispatching system includes: an input layer, a feature extraction layer, a residual block, a pooling layer and a fully connected layer;

[0010] The feature extraction of the input dispatch risk related data is performed, and then the power dispatch risk of the power dispatch system is evaluated according to the extracted features, and the power dispatch risk score corresponding to the power dispatch system is output, including:

[0011] The input layer receives the scheduling risk related data of the input model, converts the format of the scheduling risk related data into a preset format, and transmits the scheduling risk related data after the format conversion to the feature extraction layer;

[0012] The feature extraction layer extracts features from the scheduling risk related data after format conversion, and transmits the extracted features to the residual block;

[0013] The residual block performs feature processing on the feature to obtain a corresponding output feature matrix, and transmits the obtained output feature matrix to the pooling layer;

[0014] The pooling layer performs global average pooling processing on the output feature matrix, and transmits the output feature matrix after the global average pooling processing to the fully connected layer;

[0015] The fully connected layer evaluates the power dispatching risk of the power dispatching system according to the output feature matrix after global average pooling processing, and outputs the power dispatching risk score corresponding to the power dispatching system.

[0016] As a preferred solution, the generation of the risk assessment model of the power dispatching system includes:

[0017] Obtain historical dispatch risk-related data of the power dispatch system;

[0018] Performing data preprocessing on the historical scheduling risk related data; wherein the data preprocessing includes: data standardization processing, data normalization processing, data missing value processing, data outlier processing, data feature scaling processing and data encoding processing;

[0019] Taking the preprocessed historical dispatch risk related data as input and the power dispatch risk score corresponding to the historical dispatch risk related data as output, a preset ResNet neural network model is trained, and the number of the residual blocks is adjusted during the model training process to obtain the power dispatch system risk assessment model; wherein each residual block includes two convolutional layers and one skip connection.

[0020] As a preferred solution, the risk assessment model of the power dispatching system extracts features from the input dispatching risk related data, and then assesses the power dispatching risk of the power dispatching system according to the extracted features, and outputs the power dispatching risk score corresponding to the power dispatching system, and further includes:

[0021] Obtaining cross-check results of several dispatchers on the risk assessment model of the power dispatching system;

[0022] The cross-check results are incorporated into the electric power dispatching system risk assessment model as new model features, the sample weights of the electric power dispatching system risk assessment model are updated, and then the model parameters of the electric power dispatching system risk assessment model are updated according to the updated sample weights.

[0023] On the basis of the above-mentioned embodiment, another embodiment of the present invention provides a power system dispatching device based on risk assessment, comprising: a dispatching risk related data acquisition module, a power dispatching risk scoring module and a power system dispatching module;

[0024] The dispatch risk related data acquisition module is used to acquire the dispatch risk related data of the power dispatch system; wherein the dispatch risk related data includes: operation ticket data, personnel duty data and equipment data;

[0025] The power dispatching risk scoring module is used to input the dispatching risk-related data into a preset power dispatching system risk assessment model, so that the power dispatching system risk assessment model extracts features from the input dispatching risk-related data, and then assesses the power dispatching risk of the power dispatching system according to the extracted features, and outputs a power dispatching risk score corresponding to the power dispatching system; wherein the power dispatching system risk assessment model is obtained by training a preset ResNet neural network model with historical dispatching risk-related data of the power dispatching system as input and a power dispatching risk score corresponding to the historical dispatching risk-related data as output;

[0026] The power system dispatching module is used to optimize the current power dispatching strategy according to the power dispatching risk score, and then perform power system dispatching according to the optimized power dispatching strategy.

[0027] As a preferred solution, the risk assessment model of the power dispatching system includes: an input layer, a feature extraction layer, a residual block, a pooling layer and a fully connected layer;

[0028] The feature extraction of the input dispatch risk related data is performed, and then the power dispatch risk of the power dispatch system is evaluated according to the extracted features, and the power dispatch risk score corresponding to the power dispatch system is output, including:

[0029] The input layer receives the scheduling risk related data of the input model, converts the format of the scheduling risk related data into a preset format, and transmits the scheduling risk related data after the format conversion to the feature extraction layer;

[0030] The feature extraction layer extracts features from the scheduling risk related data after format conversion, and transmits the extracted features to the residual block;

[0031] The residual block performs feature processing on the feature to obtain a corresponding output feature matrix, and transmits the obtained output feature matrix to the pooling layer;

[0032] The pooling layer performs global average pooling processing on the output feature matrix, and transmits the output feature matrix after the global average pooling processing to the fully connected layer;

[0033] The fully connected layer evaluates the power dispatching risk of the power dispatching system according to the output feature matrix after global average pooling processing, and outputs the power dispatching risk score corresponding to the power dispatching system.

[0034] As a preferred solution, the generation of the risk assessment model of the power dispatching system includes:

[0035] Obtain historical dispatch risk-related data of the power dispatch system;

[0036] Performing data preprocessing on the historical scheduling risk related data; wherein the data preprocessing includes: data standardization processing, data normalization processing, data missing value processing, data outlier processing, data feature scaling processing and data encoding processing;

[0037] Taking the preprocessed historical dispatch risk related data as input and the power dispatch risk score corresponding to the historical dispatch risk related data as output, a preset ResNet neural network model is trained, and the number of the residual blocks is adjusted during the model training process to obtain the power dispatch system risk assessment model; wherein each residual block includes two convolutional layers and one skip connection.

[0038] As a preferred solution, the risk assessment model of the power dispatching system extracts features from the input dispatching risk related data, and then assesses the power dispatching risk of the power dispatching system according to the extracted features, and outputs the power dispatching risk score corresponding to the power dispatching system, and further includes:

[0039] Obtaining cross-check results of several dispatchers on the risk assessment model of the power dispatching system;

[0040] The cross-check results are incorporated into the electric power dispatching system risk assessment model as new model features, the sample weights of the electric power dispatching system risk assessment model are updated, and then the model parameters of the electric power dispatching system risk assessment model are updated according to the updated sample weights.

[0041] Based on the above embodiments, another embodiment of the present invention provides an electronic device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the risk assessment-based power system scheduling method described in the above invention embodiments is implemented.

[0042] Based on the above embodiments, another embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the risk assessment-based power system scheduling method described in the above invention embodiment.

[0043] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0044] The present invention provides a power system dispatching method based on risk assessment, which obtains dispatching risk related data of a power dispatching system; wherein the dispatching risk related data includes: operation ticket data, personnel on-duty data and equipment data; the dispatching risk related data is input into a preset power dispatching system risk assessment model, so that the power dispatching system risk assessment model extracts features from the input dispatching risk related data, and then evaluates the power dispatching risk of the power dispatching system according to the extracted features, and outputs a power dispatching risk score corresponding to the power dispatching system; according to the power dispatching risk score, the current power dispatching strategy is optimized, and then the power system is dispatched according to the optimized power dispatching strategy; wherein the power dispatching system risk assessment model is obtained by training a preset ResNet neural network model with historical dispatching risk related data of the power dispatching system as input and with the power dispatching risk score corresponding to the historical dispatching risk related data as output.

[0045] Through the present invention, the power dispatching risk of the power dispatching system can be automatically evaluated according to the dispatching risk-related data of the power dispatching system, and the power dispatching risk score corresponding to the power dispatching system can be obtained. Through the analysis and quantification of the power dispatching risk, a solid basis can be provided for the risk management of the power dispatching system and the power dispatching decision support, thereby improving the accuracy of power dispatching. Moreover, compared with the traditional dispatching method that relies on the personal experience of the dispatcher, the automated power dispatching method of the present invention improves the efficiency of power dispatching and the accuracy of operation judgment, thereby improving the safety and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flowchart of a power system dispatching method based on risk assessment provided by an embodiment of the present invention;

[0047] Figure 2 It is a schematic diagram of the construction and optimization process of the risk assessment model of the power dispatching system;

[0048] Figure 3 It is a structural diagram of the risk assessment model of the power dispatching system;

[0049] Figure 4 It is a structural schematic diagram of a power system dispatching device based on risk assessment provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned figure descriptions and any variations thereof are intended to cover non-exclusive inclusions.

[0052] In the description of the embodiments of the present application, the technical terms "first", "second", etc. are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0053] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0054] In the description of the embodiments of the present application, the term "and / or" is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0055] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0056] In the description of the embodiments of the present application, unless otherwise clearly specified and limited, technical terms such as "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of the present application can be understood according to the specific circumstances.

[0057] Embodiment 1

[0058] Please refer to Figure 1 , is a flow chart of a power system dispatching method based on risk assessment provided by an embodiment of the present invention, comprising the following specific steps:

[0059] S1. Obtaining dispatch risk related data of the power dispatch system; wherein the dispatch risk related data includes: operation ticket data, personnel duty data and equipment data;

[0060] Specifically, the present invention proposes an adaptive power dispatching operation optimization method combining human-computer interaction through innovative human-computer interaction, machine learning technology, and adaptive update mechanism, which can improve the intelligence level of power dispatching operation and ensure the safety and stability of power dispatching system operation. The specific process of the method includes:

[0061] 1. Obtaining dispatch risk related data of the power dispatch system; wherein the dispatch risk related data includes: operation ticket data, personnel duty data and equipment data;

[0062] S2. Input the dispatch risk related data into a preset power dispatch system risk assessment model, so that the power dispatch system risk assessment model extracts features from the input dispatch risk related data, and then assesses the power dispatch risk of the power dispatch system according to the extracted features, and outputs a power dispatch risk score corresponding to the power dispatch system;

[0063] 2. Input the dispatch risk related data into a pre-trained power dispatch system risk assessment model, which extracts features from the input dispatch risk related data, and then assesses the power dispatch risk of the power dispatch system based on the extracted features, and outputs a power dispatch risk score corresponding to the power dispatch system;

[0064] S3. Optimizing the current power dispatching strategy according to the power dispatching risk score, and then dispatching the power system according to the optimized power dispatching strategy;

[0065] 3. According to the power dispatch risk score, the current power dispatch strategy is optimized, and then the power system is dispatched according to the optimized power dispatch strategy.

[0066] Among them, the power dispatching system risk assessment model is obtained by training a preset ResNet neural network model with historical dispatching risk related data of the power dispatching system as input and the power dispatching risk score corresponding to the historical dispatching risk related data as output.

[0067] Preferably, the risk assessment model for the power dispatching system includes: an input layer, a feature extraction layer, a residual block, a pooling layer, and a fully connected layer; the feature extraction is performed on the input dispatching risk related data, and then the power dispatching risk of the power dispatching system is assessed according to the extracted features, and the power dispatching risk score corresponding to the power dispatching system is output, including:

[0068] The input layer receives the dispatch risk related data of the input model, converts the format of the dispatch risk related data into a preset format, and transmits the dispatch risk related data after format conversion to the feature extraction layer; the feature extraction layer extracts features from the dispatch risk related data after format conversion, and transmits the extracted features to the residual block; the residual block performs feature processing on the features to obtain a corresponding output feature matrix, and transmits the obtained output feature matrix to the pooling layer; the pooling layer performs global average pooling processing on the output feature matrix, and transmits the output feature matrix after global average pooling processing to the fully connected layer; the fully connected layer evaluates the power dispatch risk of the power dispatch system based on the output feature matrix after global average pooling processing, and outputs the power dispatch risk score corresponding to the power dispatch system.

[0069] Preferably, the generation of the risk assessment model for the electric power dispatching system includes: obtaining historical dispatching risk-related data of the electric power dispatching system; performing data preprocessing on the historical dispatching risk-related data; wherein the data preprocessing includes: data standardization processing, data normalization processing, data missing value processing, data outlier processing, data feature scaling processing and data encoding processing; taking the preprocessed historical dispatching risk-related data as input and the electric power dispatching risk score corresponding to the historical dispatching risk-related data as output, training a preset ResNet neural network model, and adjusting the number of the residual blocks during the model training process to obtain the risk assessment model for the electric power dispatching system; wherein each residual block includes two convolutional layers and one jump connection.

[0070] Specifically, for the risk assessment model of the power dispatching system, please refer to Figure 2 , is a schematic diagram of the construction and optimization process of the risk assessment model of the power dispatching system. The construction and optimization process of the risk assessment model of the power dispatching system is as follows:

[0071] 1. Human-computer interaction process: The present invention establishes a direct display interface and human-computer interaction interface. When implementing the audit, the abnormal situation is judged and adjusted through the interactive window popped up by the system. The dispatcher's experience and intuition are integrated into the decision-making process as feedback for the subsequent optimization of the machine learning model.

[0072] Feedback and iterative optimization: Collect actual operation results and dispatcher feedback, and iteratively optimize the model to improve the accuracy of risk scoring.

[0073] 2. Establishment of machine learning model: The present invention adopts the ResNet (residual network) model under the neural network model to construct the power dispatch risk assessment model, analyze the operation ticket risk data and score the risk. ResNet introduces residual modules and skip connections (Shortcut Connection), so that the gradient can be directly transferred from the back layer of the network to the front layer, effectively alleviating the gradient disappearance problem. The expressiveness and performance of the model are improved by increasing the network depth. In order to adapt to the online learning scenario of human-computer interaction, the model is modified and designed as follows.

[0074] 1. Construction of ResNet (residual network) power dispatching system risk scoring model:

[0075] Please refer to Figure 3 , is a schematic diagram of the structure of the risk assessment model for the power dispatching system, and the risk assessment model for the power dispatching system includes the following structure:

[0076] Input layer design: convert multi-source heterogeneous data, SCADA data, dispatcher operation records, etc., into a format acceptable to the model. Each data source corresponds to an input of the network.

[0077] Feature extraction: Use convolutional layers to extract features from input data. Different data sources use different convolution kernels to capture different features.

[0078] Residual block construction: Residual blocks are inserted into the network. Each residual block includes two convolutional layers and a skip connection, allowing gradients to propagate better in deep networks. Adaptive network depth is used to dynamically adjust the depth of the network according to the characteristics of the data flow and the performance of the model in a real-time environment. A shallower network is used at the beginning, and the number of residual blocks is gradually increased as the model adapts to the data. During the online learning process, the model performance is regularly evaluated to decide whether to increase or decrease the residual blocks. If the model performance does not meet expectations, continue to add residual blocks to improve the model complexity and learning ability;

[0079] Introducing the attention mechanism: Adding the self-attention mechanism can enhance the model's attention to important features, thereby improving adaptability and performance in online learning environments;

[0080] Global average pooling: Perform global average pooling on the output of the last residual block, map the high-dimensional features into vectors of fixed dimensions, and use average pooling instead of 1x1 convolution for downsampling to reduce the number of parameters and the amount of calculation;

[0081] Fully connected layer: connect the output of the pooling layer to the fully connected layer for the final risk assessment prediction;

[0082] Loss function definition: Select an appropriate loss function, such as cross entropy loss, to measure the difference between the prediction and the true label during model training.

[0083] 3. Model training and optimization process:

[0084] During the training and optimization process of the machine learning risk assessment model, the system automatically learns the operation mode and potential risk mode related to various risk parameters of power dispatching, and makes adaptive adjustments based on the feedback from dispatchers to improve the accuracy of risk assessment. The specific training and optimization process is as follows:

[0085] 1. Data preparation and interactive data collection:

[0086] (1) According to the survey conducted by the power grid dispatching department, historical dispatching risk-related data of the power dispatching system are collected and sorted, including operation ticket data, personnel duty data, equipment data, etc. The data specifically includes fields such as task type, operation steps, and historical fault records, which provide input data for risk scoring; the specific fields of each type of data are shown in Table 1 below:

[0087]

[0088] Table 1 Specific fields of historical scheduling risk related data

[0089] (2) Preprocess the data, including standardization, normalization, handling of missing values ​​and outliers, and feature scaling and encoding.

[0090] (3) The grid operation data is collected in real time through the operation ticket risk data interface. Considering that the data has time series incremental properties, the model uses the time series database InfluxDB for data storage and performs incremental processing according to the time sequence. New data will generate new risk items, which can be used to achieve continuous updating and optimization of model training.

[0091] 2. Model construction and feature extraction:

[0092] (1) Training the ResNet neural network in the task design for risk scoring task.

[0093] (2) Use convolutional layers to extract features from input data, and different data sources use different convolution kernels to capture different features.

[0094] (3) Residual blocks are inserted into the network. Each residual block consists of two convolutional layers and skip connections to improve the training efficiency of deep networks.

[0095] 3. Risk scoring model generation:

[0096] (1) Use the trained neural network model to generate a risk scoring model and quantify the risk data of the operation ticket.

[0097] (2) The model provides dispatchers with a quantitative basis for risk assessment and assists them in taking appropriate risk control measures.

[0098] 4. Online learning and interactive updating of models:

[0099] (1) Combining the output of the risk scoring model with real-time monitoring data, the training of the adaptive risk control model is realized. For the ResNet model, adaptive learning can be achieved mainly in the following ways:

[0100] Dynamically adjust residual blocks: Dynamically increase or decrease the number of residual blocks according to the complexity of the input data or the requirements of the learning task. As the data and risk types increase, the residual blocks are added to improve the learning ability of the model;

[0101] Adjust network depth: During the training process, based on the performance feedback on the validation set, gradually increase the depth of the network until a performance balance point is reached;

[0102] Structural adaptation: In addition to adjusting the number of residual blocks, other structures in the network can also be adjusted, such as adding or reducing convolutional layers, pooling layers, etc.

[0103] Parameter adaptation: Dynamically adjust network parameters to adapt to new data through online learning algorithms such as gradient descent;

[0104] Learning rate adaptation: dynamically adjust the learning rate based on the performance of the model on the validation set to optimize the learning process;

[0105] The introduction of these adaptive mechanisms enables the ResNet model to not only perform well on statically scheduled datasets, but also maintain efficiency and robustness in dynamically changing online learning environments.

[0106] Preferably, the power dispatching system risk assessment model performs feature extraction on the input dispatching risk related data, and then assesses the power dispatching risk of the power dispatching system based on the extracted features, and outputs the power dispatching risk score corresponding to the power dispatching system, and further includes: obtaining the cross-verification results of the power dispatching system risk assessment model by several dispatchers; integrating the cross-verification results into the power dispatching system risk assessment model as new model features, updating the sample weights of the power dispatching system risk assessment model, and then updating the model parameters of the power dispatching system risk assessment model according to the updated sample weights.

[0107] (2) The dispatcher interactively updates the model parameters. Considering that the model prediction results may be wrong, the dispatcher can dynamically update the model results through interaction. The system adopts the method of cross-checking between two dispatchers. When the risk adjustment suggestions are adopted by the system, the system will focus on updating the model results and further optimize the power dispatch strategy through the optimization algorithm:

[0108] Model fusion: The cross-check results of the two dispatchers are used as a new feature or signal and integrated into the model to guide model updates;

[0109] Incremental learning: Through online learning algorithms, such as variants of stochastic gradient descent (SGD), the model is allowed to gradually update weights through small batches of data, rather than training from scratch;

[0110] Feedback loop: Establish a feedback system where the dispatcher’s adjustments are used as feedback on the model performance to update the sample weights and multiply the weights of incorrect samples by 2 to fine-tune the model parameters;

[0111] Real-time monitoring: Implement a real-time monitoring system so that dispatchers can intervene and adjust immediately when model predictions deviate.

[0112] 5. Real-time monitoring and rapid response mechanism:

[0113] The real-time monitoring system can quickly identify power grid anomalies and respond quickly through adaptive control algorithms. This mechanism ensures the safety and stability of the power system and enables the dispatcher to quickly receive feedback information through technologies such as Flask service.

[0114] 6. Global average pooling and fully connected layer:

[0115] Perform global average pooling on the output of the last residual block to map high-dimensional features into vectors of fixed dimensions. Connect the output of the pooling layer to the fully connected layer for the final risk assessment prediction.

[0116] 7. Loss function and optimizer selection:

[0117] Select the loss function (cross entropy loss) and optimizer (Adam) for loss calculation and parameter update in model training.

[0118] 8. Hyperparameter tuning and regularization:

[0119] (1) Adjust hyperparameters such as learning rate, batch size, and number of hidden layer nodes to improve model performance. The model uses the above adaptive method to adjust hyperparameters.

[0120] (2) Introduce L2 regularization and Dropout technology to prevent model overfitting.

[0121] L2 regularization: L2 regularization is implemented by adding a penalty term of the L2 norm of the weight parameter to the model's loss function. This method can limit the size of the model weights and prevent the model from overfitting the training data. Mathematically, the loss function of L2 regularization can be expressed as:

[0122] L L2 =L data +λ||w|| 2 ;

[0123] Among them, L data is the data loss of the model, usually the error between the model's predicted value and the true label, such as mean square error (MSE) or cross-entropy loss. λ is the regularization parameter used to control the strength of regularization. ||w|| 2 is the square of the L2 norm of the weight vector w, expressed as the sum of the squares of the parameters in the weight vector. In the present invention, the strength of regularization can be controlled by adjusting the value of λ.

[0124] Dropout technology: Dropout is a technique that randomly "drops" neurons during training to prevent them from being overly dependent on each other's activation. In the present invention, Dropout technology can be applied in the hidden layer of the neural network. In each training iteration, each neuron has a certain probability (0.2) of being temporarily removed, which can prevent complex co-adaptation relationships between neurons and enhance the generalization ability of the model. In the test phase, in order to keep the desired output unchanged, the remaining weights are multiplied by a scaling factor Where p is the Dropout rate. The forward propagation calculation formula of Dropout can be described as: y = f(Wx + b) ⊙ m;

[0125] Where: f is the activation function, x is the input of the layer, W is the weight matrix of the layer, y is the output of the layer, and m is the Dropout mask of the layer. The probability that each element in the mask is 1 is p.

[0126] 9. Model integration and training evaluation:

[0127] The model is trained using the training data, and the validation set and test set are used to evaluate the model performance. The model is trained by stopping early if the loss of the validation set does not decrease after 10 rounds or stopping after 100 rounds of training.

[0128] 4. System-level continuous optimization and feedback loop:

[0129] The power dispatching system of the present invention implements a closed-loop optimization and feedback loop mechanism to ensure the continued accuracy and efficiency of risk assessment and operational judgment. The system achieves real-time monitoring and dynamic optimization through the following steps:

[0130] Real-time monitoring: The system monitors key performance indicators such as risk assessment accuracy, operation judgment accuracy, audit response time and lock response time in real time to ensure that all services meet established standards;

[0131] Feedback collection: Using Flask service technology, the system can quickly receive feedback from dispatchers or experts. These feedbacks are based on the rich experience of professionals and their in-depth understanding of the real-time status of power dispatching operations;

[0132] Dynamic adjustment: Based on the collected feedback, the system dynamically adjusts the risk scoring model and adaptive dispatch strategy to adapt to the changing state of the grid and emerging risk patterns;

[0133] Continuous optimization: The system is constantly iterating and updating, integrating new data and expert knowledge, optimizing model parameters and risk assessment algorithms, and improving prediction accuracy and response speed.

[0134] 5. Feedback loop of human-computer interaction and model training optimization:

[0135] This paper emphasizes the importance of human-computer interaction in the model training and optimization process, and establishes the following feedback loop mechanism:

[0136] Human-computer interaction: The dispatcher interacts with the system through an intuitive client interface, and makes judgments and adjustment suggestions on power dispatch operation data based on experience;

[0137] Feedback integration: The system collects these suggestions through the operation ticket risk data interface and the dispatcher feedback information receiving interface, and converts the results of human-computer interaction into input for model optimization;

[0138] Model retraining: The collected feedback and suggestions are used to retrain and adjust the machine learning model to ensure that the model can capture the latest risk trends and power dispatch operation characteristics;

[0139] Optimization iteration: Model optimization is not a one-time process, but a continuous iteration. The system regularly updates the model with new data and feedback to ensure the long-term stability and adaptability of the model.

[0140] It can be seen that the present invention provides a power system dispatching method based on risk assessment, which not only improves the accuracy of risk assessment and operation judgment of power dispatching, but also significantly improves the audit response efficiency and the response speed of the locking operation, thereby optimizing the operating efficiency and safety of the power dispatching monitoring system. Specifically, the present invention can achieve the following beneficial effects:

[0141] (1) First, the present invention significantly enhances the adaptability and flexibility of the power dispatching system by adopting an adaptive algorithm. This adaptability allows the system to respond to changes in the power grid state in real time and automatically adjust the dispatching strategy, thereby quickly adapting to the rapid changes in the power grid operating environment. This is unmatched by traditional systems that rely on fixed algorithms and manual experience;

[0142] (2) Secondly, the present invention uses deep learning algorithms, especially the ResNet (residual network) model, to significantly improve the intelligence level of power dispatching. This intelligence not only improves the accuracy of risk assessment, but also provides more scientific decision support for power dispatching by automatically learning the grid operation mode and potential risk mode;

[0143] (3) Furthermore, the present invention combines human-computer interaction and historical data analysis, which not only improves the efficiency of power dispatching, but also enhances the accuracy of operational judgment. Compared with traditional technologies, the present invention can process large-scale real-time data more quickly and accurately, providing powerful decision support for power dispatching;

[0144] (4) In addition, the real-time monitoring system of the present invention can quickly identify power grid anomalies and respond quickly through adaptive control algorithms. This rapid audit response and locking operation capability significantly improves the safety and stability of the power system;

[0145] (5) The present invention also provides a solid basis for risk management and decision support by quantitatively analyzing risk levels. Compared with traditional risk assessment methods, the present invention can provide more detailed and comprehensive risk assessment results, helping dispatchers make more scientific decisions;

[0146] (6) The present invention adopts an innovative machine learning model and modular system architecture design, which not only improves the flexibility and scalability of the system, but also facilitates future technology upgrades and expansions, ensuring the long-term effectiveness and adaptability of the system;

[0147] (7) The present invention also establishes a closed-loop optimization and feedback loop mechanism to ensure the continued accuracy and efficiency of risk assessment and operational judgment. Through real-time monitoring and dynamic adjustment, the system can continuously iterate and update, integrate new data and expert knowledge, optimize model parameters and risk assessment algorithms, and improve prediction accuracy and response speed.

[0148] Embodiment 2

[0149] Please refer to Figure 4 , is a schematic diagram of the structure of a power system dispatching device based on risk assessment provided by an embodiment of the present invention, the device comprising: a dispatching risk related data acquisition module, a power dispatching risk scoring module and a power system dispatching module;

[0150] The dispatch risk related data acquisition module is used to acquire the dispatch risk related data of the power dispatch system; wherein the dispatch risk related data includes: operation ticket data, personnel duty data and equipment data;

[0151] The power dispatching risk scoring module is used to input the dispatching risk-related data into a preset power dispatching system risk assessment model, so that the power dispatching system risk assessment model extracts features from the input dispatching risk-related data, and then assesses the power dispatching risk of the power dispatching system according to the extracted features, and outputs a power dispatching risk score corresponding to the power dispatching system; wherein the power dispatching system risk assessment model is obtained by training a preset ResNet neural network model with historical dispatching risk-related data of the power dispatching system as input and a power dispatching risk score corresponding to the historical dispatching risk-related data as output;

[0152] The power system dispatching module is used to optimize the current power dispatching strategy according to the power dispatching risk score, and then perform power system dispatching according to the optimized power dispatching strategy.

[0153] Preferably, the risk assessment model for the electric power dispatching system includes: an input layer, a feature extraction layer, a residual block, a pooling layer and a fully connected layer; the feature extraction is performed on the input dispatching risk-related data, and then the electric power dispatching risk of the electric power dispatching system is evaluated according to the extracted features, and the electric power dispatching risk score corresponding to the electric power dispatching system is output, including: the input layer receives the dispatching risk-related data of the input model, converts the format of the dispatching risk-related data into a preset format, and transmits the format-converted dispatching risk-related data to the feature extraction layer; the feature extraction layer extracts features from the format-converted dispatching risk-related data, and transmits the extracted features to the residual block; the residual block performs feature processing on the features to obtain the corresponding output feature matrix, and transmits the obtained output feature matrix to the pooling layer; the pooling layer performs global average pooling processing on the output feature matrix, and transmits the output feature matrix after the global average pooling processing to the fully connected layer; the fully connected layer evaluates the electric power dispatching risk of the electric power dispatching system according to the output feature matrix after the global average pooling processing, and outputs the electric power dispatching risk score corresponding to the electric power dispatching system.

[0154] Preferably, the generation of the risk assessment model for the electric power dispatching system includes: obtaining historical dispatching risk-related data of the electric power dispatching system; performing data preprocessing on the historical dispatching risk-related data; wherein the data preprocessing includes: data standardization processing, data normalization processing, data missing value processing, data outlier processing, data feature scaling processing and data encoding processing; taking the preprocessed historical dispatching risk-related data as input and the electric power dispatching risk score corresponding to the historical dispatching risk-related data as output, training a preset ResNet neural network model, and adjusting the number of the residual blocks during the model training process to obtain the risk assessment model for the electric power dispatching system; wherein each residual block includes two convolutional layers and one jump connection.

[0155] Preferably, the power dispatching system risk assessment model performs feature extraction on the input dispatching risk related data, and then assesses the power dispatching risk of the power dispatching system based on the extracted features, and outputs the power dispatching risk score corresponding to the power dispatching system, and further includes: obtaining the cross-verification results of the power dispatching system risk assessment model by several dispatchers; integrating the cross-verification results into the power dispatching system risk assessment model as new model features, updating the sample weights of the power dispatching system risk assessment model, and then updating the model parameters of the power dispatching system risk assessment model according to the updated sample weights.

[0156] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0157] Those skilled in the art can clearly understand that, for the sake of convenience and brevity, the specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0158] Embodiment 3

[0159] Accordingly, an embodiment of the present invention provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the risk assessment-based power system scheduling method described in the above-mentioned embodiment of the invention is implemented.

[0160] The electronic device may be a computing device such as a desktop computer, a notebook, a palm computer, a cloud server, etc. The device may include, but is not limited to, a processor and a memory.

[0161] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the device, and various interfaces and lines are used to connect various parts of the entire device.

[0162] Embodiment 4

[0163] Accordingly, an embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the risk assessment-based power system scheduling method described in the above-mentioned embodiment of the invention.

[0164] The memory can be used to store the computer program, and the processor realizes various functions of the device by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0165] The storage medium is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When the computer program is executed by the processor, the steps of each method embodiment described above can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0166] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A power system dispatching method based on risk assessment, characterized in that: include: Obtaining dispatch risk related data of the power dispatch system; wherein the dispatch risk related data includes: operation ticket data, personnel duty data and equipment data; Inputting the dispatch risk related data into a preset power dispatch system risk assessment model, so that the power dispatch system risk assessment model extracts features from the input dispatch risk related data, and then assesses the power dispatch risk of the power dispatch system according to the extracted features, and outputs a power dispatch risk score corresponding to the power dispatch system; Optimizing the current power dispatching strategy according to the power dispatching risk score, and then dispatching the power system according to the optimized power dispatching strategy; Among them, the power dispatching system risk assessment model is obtained by training a preset ResNet neural network model with historical dispatching risk related data of the power dispatching system as input and the power dispatching risk score corresponding to the historical dispatching risk related data as output.

2. The power system dispatching method based on risk assessment according to claim 1, characterized in that: The electric power dispatching system risk assessment model comprises: an input layer, a feature extraction layer, a residual block, a pooling layer and a fully connected layer; The feature extraction of the input dispatch risk related data is performed, and then the power dispatch risk of the power dispatch system is evaluated according to the extracted features, and the power dispatch risk score corresponding to the power dispatch system is output, including: The input layer receives the scheduling risk related data of the input model, converts the format of the scheduling risk related data into a preset format, and transmits the scheduling risk related data after the format conversion to the feature extraction layer; The feature extraction layer extracts features from the scheduling risk related data after format conversion, and transmits the extracted features to the residual block; The residual block performs feature processing on the feature to obtain a corresponding output feature matrix, and transmits the obtained output feature matrix to the pooling layer; The pooling layer performs global average pooling processing on the output feature matrix, and transmits the output feature matrix after the global average pooling processing to the fully connected layer; The fully connected layer evaluates the power dispatching risk of the power dispatching system according to the output feature matrix after global average pooling processing, and outputs the power dispatching risk score corresponding to the power dispatching system.

3. The power system dispatching method based on risk assessment according to claim 2, characterized in that: The generation of the risk assessment model of the power dispatching system includes: Obtain historical dispatch risk-related data of the power dispatch system; Performing data preprocessing on the historical scheduling risk related data; wherein the data preprocessing includes: data standardization processing, data normalization processing, data missing value processing, data outlier processing, data feature scaling processing and data encoding processing; Taking the preprocessed historical dispatch risk related data as input and the power dispatch risk score corresponding to the historical dispatch risk related data as output, a preset ResNet neural network model is trained, and the number of the residual blocks is adjusted during the model training process to obtain the power dispatch system risk assessment model; wherein each residual block includes two convolutional layers and one skip connection.

4. The power system dispatching method based on risk assessment according to claim 1, characterized in that: The power dispatching system risk assessment model extracts features from the input dispatching risk related data, and then assesses the power dispatching risk of the power dispatching system based on the extracted features, and outputs the power dispatching risk score corresponding to the power dispatching system, and also includes: Obtaining cross-check results of several dispatchers on the risk assessment model of the power dispatching system; The cross-check results are incorporated into the electric power dispatching system risk assessment model as new model features, the sample weights of the electric power dispatching system risk assessment model are updated, and then the model parameters of the electric power dispatching system risk assessment model are updated according to the updated sample weights.

5. A power system dispatching device based on risk assessment, characterized in that: include: Dispatch risk related data acquisition module, power dispatch risk scoring module and power system dispatch module; The dispatch risk related data acquisition module is used to acquire the dispatch risk related data of the power dispatch system; wherein the dispatch risk related data includes: operation ticket data, personnel duty data and equipment data; The power dispatching risk scoring module is used to input the dispatching risk-related data into a preset power dispatching system risk assessment model, so that the power dispatching system risk assessment model extracts features from the input dispatching risk-related data, and then assesses the power dispatching risk of the power dispatching system according to the extracted features, and outputs a power dispatching risk score corresponding to the power dispatching system; wherein the power dispatching system risk assessment model is obtained by training a preset ResNet neural network model with historical dispatching risk-related data of the power dispatching system as input and a power dispatching risk score corresponding to the historical dispatching risk-related data as output; The power system dispatching module is used to optimize the current power dispatching strategy according to the power dispatching risk score, and then perform power system dispatching according to the optimized power dispatching strategy.

6. The power system dispatching device based on risk assessment according to claim 5, characterized in that: The electric power dispatching system risk assessment model comprises: an input layer, a feature extraction layer, a residual block, a pooling layer and a fully connected layer; The feature extraction of the input dispatch risk related data is performed, and then the power dispatch risk of the power dispatch system is evaluated according to the extracted features, and the power dispatch risk score corresponding to the power dispatch system is output, including: The input layer receives the scheduling risk related data of the input model, converts the format of the scheduling risk related data into a preset format, and transmits the scheduling risk related data after the format conversion to the feature extraction layer; The feature extraction layer extracts features from the scheduling risk related data after format conversion, and transmits the extracted features to the residual block; The residual block performs feature processing on the feature to obtain a corresponding output feature matrix, and transmits the obtained output feature matrix to the pooling layer; The pooling layer performs global average pooling processing on the output feature matrix, and transmits the output feature matrix after the global average pooling processing to the fully connected layer; The fully connected layer evaluates the power dispatching risk of the power dispatching system according to the output feature matrix after global average pooling processing, and outputs the power dispatching risk score corresponding to the power dispatching system.

7. The power system dispatching device based on risk assessment according to claim 6, characterized in that: The generation of the risk assessment model of the power dispatching system includes: Obtain historical dispatch risk-related data of the power dispatch system; Performing data preprocessing on the historical scheduling risk related data; wherein the data preprocessing includes: data standardization processing, data normalization processing, data missing value processing, data outlier processing, data feature scaling processing and data encoding processing; Taking the preprocessed historical dispatch risk related data as input and the power dispatch risk score corresponding to the historical dispatch risk related data as output, a preset ResNet neural network model is trained, and the number of the residual blocks is adjusted during the model training process to obtain the power dispatch system risk assessment model; wherein each residual block includes two convolutional layers and one skip connection.

8. The power system dispatching device based on risk assessment according to claim 5, characterized in that: The power dispatching system risk assessment model extracts features from the input dispatching risk related data, and then assesses the power dispatching risk of the power dispatching system based on the extracted features, and outputs the power dispatching risk score corresponding to the power dispatching system, and also includes: Obtaining cross-check results of several dispatchers on the risk assessment model of the power dispatching system; The cross-check results are incorporated into the electric power dispatching system risk assessment model as new model features, the sample weights of the electric power dispatching system risk assessment model are updated, and then the model parameters of the electric power dispatching system risk assessment model are updated according to the updated sample weights.

9. An electronic device, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements the risk assessment-based power system dispatching method as described in any one of claims 1 to 4.

10. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is executed, the device where the storage medium is located is controlled to execute the risk assessment-based power system dispatching method according to any one of claims 1 to 4.

Images