A scheduling optimization method and system based on distributed measurement and control

Abstract

The present application discloses a scheduling optimization method and system based on distributed measurement and control, including: inputting real-time node data into an anomaly detection model for anomaly detection, and scheduling the object to be scheduled according to the anomaly detection result; the training process of the anomaly detection model includes: determining a first data type and a first standard feature, a second data type and a second standard feature; extracting target node data from sample data according to the first data type, and extracting data features of the target node data through a model to be trained; determining a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reducing the second difference coefficient to obtain a third difference coefficient; determining a model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient; and training the model to be trained according to the model adjustment parameter to obtain an anomaly detection model to improve the accuracy of scheduling.

Description

Technical Field

[0001] The present application relates to the field of scheduling technology, and in particular to a scheduling optimization method and system based on distributed measurement and control. Background Art

[0002] At present, the types and quantities of power equipment connected to the power network are increasing. In order to improve the accuracy of scheduling, it is usually necessary to set corresponding sensors in various power equipment, and then hand over the sensor data to the scheduling end as the basis for scheduling.

[0003] However, in a power grid to be dispatched, the total amount of sensor data is very large, and it is difficult for the dispatcher to accurately identify abnormal sensor data, and further difficult to identify the power equipment that needs to be dispatched, making the dispatching inaccurate. Summary of the invention

[0004] In order to solve the above technical problems, the embodiments of the present application propose a scheduling optimization method and system based on distributed measurement and control, which can improve the accuracy of scheduling.

[0005] In a first aspect, an embodiment of the present application provides a scheduling optimization method based on distributed measurement and control, which is applicable to a cloud connected to a distributed measurement and control node for communication, wherein the distributed measurement and control node is used to detect an object to be scheduled to generate real-time node data and transmit the real-time node data to the cloud, and the method includes:

[0006] The real-time node data is input into a pre-trained anomaly detection model for anomaly detection, so as to schedule the object to be scheduled according to the anomaly detection result.

[0007] The training process of the anomaly detection model includes:

[0008] A first data type and a first standard feature corresponding thereto, a second data type and a second standard feature corresponding thereto are determined respectively, wherein the first data type is different from the second data type.

[0009] Target node data is extracted from the sample data according to the first data type, and data features of the target node data are extracted through the model to be trained.

[0010] Determine a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reduce the second difference coefficient to obtain a third difference coefficient.

[0011] A model adjustment parameter is determined based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

[0012] The model to be trained is trained according to the model adjustment parameters to obtain the anomaly detection model.

[0013] Furthermore, determining the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes:

[0014] A first similarity is determined based on the data feature, the first standard feature, and the first difference coefficient.

[0015] A second similarity is determined based on the data feature, the second standard feature and the third difference coefficient.

[0016] The model adjustment parameter is determined based on the first similarity and the second similarity.

[0017] Furthermore, determining the model adjustment parameter based on the first similarity and the second similarity includes:

[0018] A sum of the first similarity and the second similarity is calculated.

[0019] The model adjustment parameter is determined based on a ratio between the first similarity and the sum value.

[0020] Furthermore, the method further comprises:

[0021] Determine the data volume ratio of the target node data in the sample data to determine the feature weight according to the data volume ratio.

[0022] The determining of the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes:

[0023] A model adjustment parameter is determined based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

[0024] Furthermore, determining the model adjustment parameter based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes:

[0025] A first similarity is determined based on the data feature, the first standard feature, and the first difference coefficient.

[0026] A second similarity is determined based on the data feature, the second standard feature and the third difference coefficient.

[0027] The model adjustment parameter is determined based on the feature weight, the first similarity, and the second similarity.

[0028] Furthermore, the first similarity is determined by the product of the norm of the data feature, the norm of the first standard feature, and the first difference coefficient.

[0029] The second similarity is determined by the product of the norm of the data feature, the norm of the second standard feature, and the third difference coefficient.

[0030] Furthermore, the method further comprises:

[0031] A scaling factor is determined based on the first difference coefficient.

[0032] The step of reducing the second difference coefficient to obtain a third difference coefficient includes:

[0033] The third difference coefficient is obtained by multiplying the second difference coefficient by the scaling ratio.

[0034] Furthermore, the first difference coefficient is used to characterize the degree of deviation between the data feature and the first standard feature, wherein the smaller the degree of deviation is, the smaller the scaling ratio is.

[0035] Furthermore, the objects to be dispatched include various electric power equipment in the power grid to be dispatched.

[0036] In a second aspect, an embodiment of the present application provides a scheduling optimization system based on distributed measurement and control, including a distributed measurement and control node and a cloud connected to the distributed measurement and control node in communication, wherein the distributed measurement and control node is used to detect an object to be scheduled to generate real-time node data and transmit the real-time node data to the cloud, and the cloud is configured as follows:

[0037] The real-time node data is input into a pre-trained anomaly detection model for anomaly detection, so as to schedule the object to be scheduled according to the anomaly detection result.

[0038] The training process of the anomaly detection model includes:

[0039] A first data type and a first standard feature corresponding thereto, a second data type and a second standard feature corresponding thereto are determined respectively, wherein the first data type is different from the second data type.

[0040] Target node data is extracted from the sample data according to the first data type, and data features of the target node data are extracted through the model to be trained.

[0041] Determine a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reduce the second difference coefficient to obtain a third difference coefficient.

[0042] A model adjustment parameter is determined based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

[0043] The model to be trained is trained according to the model adjustment parameters to obtain the anomaly detection model.

[0044] In summary, the embodiments of the present application have at least the following beneficial effects:

[0045] According to an embodiment of the present application, the real-time node data is input into a pre-trained anomaly detection model for anomaly detection, so that the object to be scheduled is scheduled according to the anomaly detection result; wherein the training process of the anomaly detection model includes: respectively determining a first data type and its corresponding first standard feature, a second data type and its corresponding second standard feature, wherein the first data type is different from the second data type; extracting target node data from sample data according to the first data type, and extracting data features of the target node data through the model to be trained; determining a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reducing the second difference coefficient to obtain a third difference coefficient; determining a model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient; training the model to be trained according to the model adjustment parameter to obtain the anomaly detection model, so that the anomaly detection model can improve the recognition ability of different types of data, that is, improve the ability to distinguish normal data from abnormal data, and then the model can be used to quickly and accurately determine whether the real-time node data is abnormal, thereby improving the accuracy of scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of a scheduling optimization method based on distributed measurement and control provided in an embodiment of the present application;

[0047] Figure 2 It is a flowchart of the training process of the anomaly detection model provided in the embodiment of the present application;

[0048] Figure 3 It is a structural diagram of a scheduling optimization system based on distributed measurement and control provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0050] In the description of the present application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, features defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "multiple" is two or more. In the description of the present application, the term "including" and its variations are open inclusions, i.e., "including but not limited to". The term "based on" means "at least partially based on". The term "according to" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments".

[0051] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" 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 it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0052] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0053] First, see Figure 1 , showing a flow chart of a scheduling optimization method based on distributed measurement and control provided in an embodiment of the present application, the method is applicable to a cloud that is communicatively connected to a distributed measurement and control node, the distributed measurement and control node is used to detect an object to be scheduled to generate real-time node data and transmit the real-time node data to the cloud, the method includes step S101, which is specifically as follows.

[0054] S101, inputting the real-time node data into a pre-trained anomaly detection model to perform anomaly detection, so as to schedule the object to be scheduled according to the anomaly detection result.

[0055] It should be noted that the distributed measurement and control nodes in this embodiment refer to multiple measurement and control units dispersedly arranged in the network, which are used to collect data and transmit it to the cloud through the network so that the cloud can perform real-time monitoring and control. In the power system, these nodes can include various sensors, smart meters, circuit breaker controllers, and other types of smart devices, which can be distributed in a wide geographical area and together constitute a complex monitoring and control system.

[0056] Distributed measurement and control nodes can have the following characteristics:

[0057] Distribution: Nodes are distributed throughout the power grid, covering multiple locations such as power stations, substations, transmission lines, and user terminals.

[0058] Autonomy: Each node has a certain degree of autonomous processing capability and can independently complete data collection, preliminary analysis and local control tasks to a certain extent.

[0059] Communication capability: Nodes can be connected to each other through wired or wireless networks to achieve data exchange and collaborative work.

[0060] Reliability: The failure of a single node will not affect the operation of the entire system, because other nodes can continue to work or take over the tasks of the failed node.

[0061] Among them, see Figure 2 , shows a flow chart of a training process of an anomaly detection model provided in an embodiment of the present application, including steps S201-S205, as follows.

[0062] S201, respectively determine a first data type and a first standard feature corresponding thereto, a second data type and a second standard feature corresponding thereto, wherein the first data type is different from the second data type.

[0063] It can be understood that in this embodiment, the first data type can be used to characterize normal data, and the second data type can be used to characterize abnormal data. For ease of explanation, this case is used below for illustration; but on the other hand, it should be understood that the first data type can also be used to characterize abnormal data, and the second data type is used to characterize normal data. In addition, the standard feature refers to a pre-configured data feature of the corresponding data type, such as a feature vector composed of the standard values ​​of its mean, variance, and peak value.

[0064] S202: extract target node data from sample data according to the first data type, and extract data features of the target node data through the model to be trained.

[0065] It can be understood that the target node data in this embodiment is data extracted from the data belonging to the first data type in the sample data. The model to be trained can extract data features from the target node data, and the data features and each feature in the embodiment of the present application can be represented by a feature vector.

[0066] S203, determining a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reducing the second difference coefficient to obtain a third difference coefficient.

[0067] In one example, determining a first difference coefficient between the data feature and the first standard feature and a second difference coefficient between the data feature and the second standard feature may include:

[0068] A measurement method (such as Euclidean distance and Manhattan distance) is used to calculate the first difference between the data feature and the first standard feature, and the first difference is used as the first difference coefficient.

[0069] A measurement method (such as Euclidean distance and Manhattan distance) is used to calculate a second difference between the data feature and the second standard feature, and the second difference is used as a second difference coefficient.

[0070] In one example, determining a first difference coefficient between the data feature and the first standard feature and a second difference coefficient between the data feature and the second standard feature may include:

[0071] A first angle between a data feature vector indicated by the data feature and a first standard feature vector indicated by the first standard feature is determined, and a cosine value of the first angle is calculated as a first difference coefficient.

[0072] A second angle between a data feature vector indicated by the data feature and a second standard feature vector indicated by the second standard feature is determined, and a cosine value of the second angle is calculated as a second difference coefficient.

[0073] S204: Determine a model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

[0074] In one example, determining a model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient may include:

[0075] The data features, the first standard features, the second standard features, the first difference coefficient and the third difference coefficient are input into a pre-trained adjustment parameter generation model to obtain model adjustment parameters generated by the adjustment parameter generation model based on the data features, the first standard features, the second standard features, the first difference coefficient and the third difference coefficient.

[0076] Among them, the training method of the adjustment parameter generation model may include 1)-4):

[0077] 1) Data preparation.

[0078] Collect data: First, you need to collect a large number of data sets containing data features, first standard features, second standard features, first difference coefficients, third difference coefficients, and corresponding optimal or expected model adjustment parameters. These data can come from historical experiments, simulation data, or expert annotations.

[0079] Preprocessing: Cleaning, denoising, and normalizing or standardizing the collected data to ensure that the model can learn more effectively.

[0080] Feature engineering: Design appropriate feature representation based on business requirements and data characteristics, which may include feature selection, feature extraction or feature transformation.

[0081] 2) Model design.

[0082] Choose a model type: Choose an appropriate model type based on the complexity of the problem and the characteristics of the data. For this type of problem, you can consider using a regression model (such as linear regression, decision tree regression, random forest regression, GBDT (Gradient Boosting Decision Tree), neural network, etc.) because the target (model adjustment parameter) is a continuous value.

[0083] Model architecture: If you choose a neural network, you need to design the network structure, including the number of layers, the number of neurons in each layer, the activation function, the loss function, etc.

[0084] 3) Model training.

[0085] Divide the dataset: Divide the dataset into training set, validation set and test set, usually in the proportion of 70%, 15%, 15% or similar.

[0086] Training model: Use the training set data to train the model and adjust the model parameters through optimization methods such as back propagation algorithm to minimize the difference between the predicted value and the actual value (i.e., the loss function).

[0087] Validation and parameter adjustment: Use validation set data to evaluate model performance and adjust parameters (such as learning rate, batch size, number of iterations, etc.) as needed to find the optimal model configuration.

[0088] 4) Model evaluation.

[0089] Performance evaluation: Use the test set data to evaluate the performance of the final model to ensure that the model has good generalization ability. Evaluation indicators can be mean square error, mean absolute error, R^2 score, etc.

[0090] Error analysis: Analyze the model’s errors on the test set, identify possible deviations or deficiencies, and provide a basis for subsequent model improvements.

[0091] S205: Train the model to be trained according to the model adjustment parameters to obtain the anomaly detection model.

[0092] It can be understood that in this embodiment, the model parameters of the model to be trained can be adjusted directly according to the model adjustment parameters to achieve the training effect.

[0093] In an optional implementation, determining the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient includes:

[0094] A first similarity is determined based on the data feature, the first standard feature, and the first difference coefficient.

[0095] It should be noted that the first similarity in this embodiment can be used to characterize the similarity between the data feature and the first standard feature. In some examples, the similarity between the data feature and the first standard feature can be characterized by calculating the cosine similarity, Euclidean distance, etc. between the data feature and the first standard feature. In this way, the first difference coefficient can be used as an adjustment coefficient to adjust the similarity between the data feature and the first standard feature to obtain the first similarity. For example, the two can be multiplied to achieve adjustment.

[0096] A second similarity is determined based on the data feature, the second standard feature and the third difference coefficient.

[0097] It should be noted that the first similarity in this embodiment can be used to characterize the similarity between the data feature and the second standard feature. In some examples, the similarity between the data feature and the second standard feature can be characterized by calculating the cosine similarity, Euclidean distance, etc. between the data feature and the second standard feature. In this way, the second difference coefficient can be used as an adjustment coefficient to adjust the similarity between the data feature and the second standard feature to obtain the second similarity. For example, the two can be multiplied to achieve adjustment.

[0098] In this embodiment, since the second difference coefficient can characterize the difference between the data feature and the second standard feature, when the third difference coefficient that is smaller than the second difference coefficient is applied to the determination of the second similarity, the obtained second similarity corresponds to the feature information after reducing the difference between the data feature and the second standard feature, so as to improve the model's ability to distinguish between the second standard feature and the data feature, thereby improving the model's ability to recognize the data type to which the node data belongs.

[0099] The model adjustment parameter is determined based on the first similarity and the second similarity.

[0100] In one example, determining the model adjustment parameter based on the first similarity and the second similarity may include: adjusting the model parameter using an optimization algorithm (eg, gradient descent method) based on the first similarity and the second similarity to determine the optimal model adjustment parameter.

[0101] In an optional implementation, determining the model adjustment parameter based on the first similarity and the second similarity includes:

[0102] A sum of the first similarity and the second similarity is calculated.

[0103] The model adjustment parameter is determined based on a ratio between the first similarity and the sum value.

[0104] In one example, the model adjusts the parameters It can be expressed by the following formula:

[0105]

[0106] in, represents the first similarity, represents the second similarity, is the natural base, is the natural logarithm, It represents the product of the norm of the data feature, the norm of the first standard feature, and the first coefficient of variation. It represents the product of the norm of the data feature, the norm of the second standard feature, and the third coefficient of variation.

[0107] In an optional embodiment, the method further includes:

[0108] Determine the data volume ratio of the target node data in the sample data to determine the feature weight according to the data volume ratio.

[0109] It is understandable that since it is difficult to obtain different types of sample data, the number of samples of different types is also different. At this time, for types with a small number of samples, the model may not be able to fully learn the characteristics of this type, resulting in poor generalization of the model. Therefore, in one example, there is a negative correlation between the data volume ratio and the feature weight, that is, the larger the data volume ratio, the smaller the feature weight, and the smaller the data volume ratio, the larger the feature weight, so that the model can fully learn the characteristics of the type with a small data volume ratio.

[0110] The determining of the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes:

[0111] A model adjustment parameter is determined based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

[0112] It can be understood that the model adjustment parameters of this embodiment can also be generated using the above-mentioned adjustment parameter generation model. It is only necessary to set the corresponding weight coefficients for the adjustment parameter generation model, which will not be described in detail here.

[0113] In an optional implementation, determining the model adjustment parameter based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient includes:

[0114] A first similarity is determined based on the data feature, the first standard feature, and the first difference coefficient.

[0115] A second similarity is determined based on the data feature, the second standard feature and the third difference coefficient.

[0116] The model adjustment parameter is determined based on the feature weight, the first similarity, and the second similarity.

[0117] In one example, the model adjusts the parameters It can be expressed by the following formula:

[0118]

[0119] in, represents the first similarity, represents the second similarity, is the natural base, is the natural logarithm, It represents the product of the norm of the data feature, the norm of the first standard feature, and the first coefficient of variation. The product of the norm of the data feature, the norm of the second standard feature, and the third coefficient of variation, Represents feature weight.

[0120] In an optional implementation, the first similarity is determined by the product of the norm of the data feature, the norm of the first standard feature, and the first difference coefficient.

[0121] In one example, the first similarity can be ,in, Represents the product of the norm of the data feature, the norm of the first standard feature, and the first coefficient of variation.

[0122] The second similarity is determined by the product of the norm of the data feature, the norm of the second standard feature, and the third difference coefficient.

[0123] In one example, the first similarity can be ,in, It represents the product of the norm of the data feature, the norm of the second standard feature, and the third coefficient of variation.

[0124] In an optional embodiment, the method further includes:

[0125] A scaling factor is determined based on the first difference coefficient.

[0126] The step of reducing the second difference coefficient to obtain a third difference coefficient includes:

[0127] The third difference coefficient is obtained by multiplying the second difference coefficient by the scaling ratio.

[0128] In an optional implementation, the first difference coefficient is used to characterize the degree of deviation between the data feature and the first standard feature, wherein the smaller the degree of deviation is, the smaller the scaling ratio is.

[0129] In an optional implementation, the objects to be scheduled include various power equipment in the power grid to be scheduled.

[0130] For example, in the power system, the power equipment in the power grid to be dispatched mainly includes power generation equipment, transmission equipment, distribution equipment, energy storage equipment and other auxiliary equipment, etc. These equipment and their roles in power grid dispatch are described in detail below.

[0131] 1) Power generation equipment.

[0132] Thermal power plant: uses fossil fuels (such as coal, natural gas, etc.) to generate heat energy to drive steam turbines to rotate and generate electricity.

[0133] Hydroelectric power station: uses the kinetic energy generated by water flow to drive the turbine to rotate and generate electricity.

[0134] Nuclear power station: uses the heat energy generated by nuclear reactions to drive the turbine to rotate and generate electricity.

[0135] Renewable energy power generation: including wind power generation, solar power generation, biomass power generation, etc.

[0136] 2) Power transmission equipment.

[0137] Transmission Lines: High voltage transmission lines are used to transmit electricity over long distances.

[0138] Transformer: used to transform voltage levels to meet the transmission and distribution requirements of different voltage levels.

[0139] Circuit breaker: Used to protect circuits from overload or short circuit and can quickly cut off fault current.

[0140] 3) Power distribution equipment.

[0141] Distribution substations: Convert high voltage electricity to lower voltage suitable for domestic and industrial use.

[0142] Distribution lines: used to transmit electricity from distribution stations to users.

[0143] Smart meter: used to monitor users' electricity usage and support functions such as remote meter reading.

[0144] 4) Energy storage equipment.

[0145] Battery energy storage system: uses battery technology to store electrical energy for future use.

[0146] Pumped storage power station: uses the potential energy of water to store electrical energy and generates electricity by releasing water when needed.

[0147] Supercapacitors: They have the ability to charge and discharge quickly and can be used for short-term energy storage.

[0148] 5) Auxiliary equipment.

[0149] Reactive power compensation device: used to improve the power factor of the power grid and reduce power loss.

[0150] Phase regulator: used to regulate reactive power in the power grid and maintain voltage level.

[0151] Relay protection devices: Used to detect faults in the power grid and take prompt action to isolate the faulty area.

[0152] Optionally, the real-time node data in the embodiments of the present application can be transmitted via blockchain technology to ensure the security of the data.

[0153] Second, see Figure 3 , shows a schematic diagram of the structure of a scheduling optimization system based on distributed measurement and control provided in an embodiment of the present application, the scheduling optimization system based on distributed measurement and control includes a distributed measurement and control node 301 and a cloud 302 that is communicatively connected to the distributed measurement and control node, the distributed measurement and control node 301 is used to detect the object to be scheduled to generate real-time node data and transmit the real-time node data to the cloud 302, the cloud 302 is configured as:

[0154] The real-time node data is input into a pre-trained anomaly detection model for anomaly detection, so as to schedule the object to be scheduled according to the anomaly detection result.

[0155] The training process of the anomaly detection model includes:

[0156] A first data type and a first standard feature corresponding thereto, a second data type and a second standard feature corresponding thereto are determined respectively, wherein the first data type is different from the second data type.

[0157] Target node data is extracted from the sample data according to the first data type, and data features of the target node data are extracted through the model to be trained.

[0158] Determine a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reduce the second difference coefficient to obtain a third difference coefficient.

[0159] A model adjustment parameter is determined based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

[0160] The model to be trained is trained according to the model adjustment parameters to obtain the anomaly detection model.

[0161] In an optional implementation, determining the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient includes:

[0162] A first similarity is determined based on the data feature, the first standard feature, and the first difference coefficient.

[0163] A second similarity is determined based on the data feature, the second standard feature and the third difference coefficient.

[0164] The model adjustment parameter is determined based on the first similarity and the second similarity.

[0165] In an optional implementation, determining the model adjustment parameter based on the first similarity and the second similarity includes:

[0166] A sum of the first similarity and the second similarity is calculated.

[0167] The model adjustment parameter is determined based on a ratio between the first similarity and the sum value.

[0168] In an optional implementation, the cloud is further configured as:

[0169] Determine the data volume ratio of the target node data in the sample data to determine the feature weight according to the data volume ratio.

[0170] The determining of the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes:

[0171] A model adjustment parameter is determined based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

[0172] In an optional implementation, determining the model adjustment parameter based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient includes:

[0173] A first similarity is determined based on the data feature, the first standard feature, and the first difference coefficient.

[0174] A second similarity is determined based on the data feature, the second standard feature and the third difference coefficient.

[0175] The model adjustment parameter is determined based on the feature weight, the first similarity, and the second similarity.

[0176] In an optional implementation, the first similarity is determined by the product of the norm of the data feature, the norm of the first standard feature, and the first difference coefficient.

[0177] The second similarity is determined by the product of the norm of the data feature, the norm of the second standard feature, and the third difference coefficient.

[0178] In an optional implementation, the cloud is further configured as:

[0179] A scaling factor is determined based on the first difference coefficient.

[0180] The step of reducing the second difference coefficient to obtain a third difference coefficient includes:

[0181] The third difference coefficient is obtained by multiplying the second difference coefficient by the scaling ratio.

[0182] In an optional implementation, the first difference coefficient is used to characterize the degree of deviation between the data feature and the first standard feature, wherein the smaller the degree of deviation is, the smaller the scaling ratio is.

[0183] In an optional implementation, the objects to be scheduled include various power equipment in the power grid to be scheduled.

[0184] In summary, the embodiments of the present application have at least the following beneficial effects:

[0185] According to an embodiment of the present application, the real-time node data is input into a pre-trained anomaly detection model for anomaly detection, so that the object to be scheduled is scheduled according to the anomaly detection result; wherein the training process of the anomaly detection model includes: respectively determining a first data type and its corresponding first standard feature, a second data type and its corresponding second standard feature, wherein the first data type is different from the second data type; extracting target node data from sample data according to the first data type, and extracting data features of the target node data through the model to be trained; determining a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reducing the second difference coefficient to obtain a third difference coefficient; determining a model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient; training the model to be trained according to the model adjustment parameter to obtain the anomaly detection model, so that the anomaly detection model can improve the recognition ability of different types of data, that is, improve the ability to distinguish normal data from abnormal data, and then the model can be used to quickly and accurately determine whether the real-time node data is abnormal, thereby improving the accuracy of scheduling.

[0186] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary hardware platform, and of course, it can also be implemented entirely by hardware. Based on such an understanding, all or part of the contribution of the technical solution of the present application to the background technology can be embodied in the form of a software product, and the computer software product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application or some parts of the embodiments.

[0187] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications are also considered to be within the scope of protection of the present application.

Claims

1. A scheduling optimization method based on distributed measurement and control, characterized in that: Applicable to a cloud connected to a distributed measurement and control node for detecting an object to be scheduled to generate real-time node data and transmitting the real-time node data to the cloud, the method comprising: Inputting the real-time node data into a pre-trained anomaly detection model for anomaly detection, so as to schedule the object to be scheduled according to the anomaly detection result; The training process of the anomaly detection model includes: respectively determining a first data type and a first standard feature corresponding thereto, and a second data type and a second standard feature corresponding thereto, wherein the first data type is different from the second data type; Extracting target node data from sample data according to the first data type, and extracting data features of the target node data through the model to be trained; Determine a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reduce the second difference coefficient to obtain a third difference coefficient; Determining a model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient; Train the model to be trained according to the model adjustment parameters to obtain the anomaly detection model; Wherein, determining the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes: Determining a first similarity based on the data feature, the first standard feature, and the first difference coefficient; Determining a second similarity based on the data feature, the second standard feature, and the third difference coefficient; Determining the model adjustment parameter based on the first similarity and the second similarity; The step of determining the model adjustment parameter based on the first similarity and the second similarity includes: Calculating a sum of the first similarity and the second similarity; Based on the ratio between the first similarity and the sum value, the model adjustment parameter is determined. It is expressed by the following formula: in, represents the first similarity, represents the second similarity, is the natural base, is the natural logarithm, It represents the product of the norm of the data feature, the norm of the first standard feature, and the first coefficient of variation. It represents the product of the norm of the data feature, the norm of the second standard feature, and the third coefficient of variation; The objects to be dispatched include various electric power equipment in the power grid to be dispatched.

2. The scheduling optimization method based on distributed measurement and control according to claim 1, characterized in that: The method further comprises: Determine the data volume ratio of the target node data in the sample data, and determine the feature weight according to the data volume ratio; The determining of the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient also includes: A model adjustment parameter is determined based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient.

3. The scheduling optimization method based on distributed measurement and control as claimed in claim 2, characterized in that: The determining of the model adjustment parameter based on the feature weight, the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes: Determining a first similarity based on the data feature, the first standard feature, and the first difference coefficient; Determining a second similarity based on the data feature, the second standard feature, and the third difference coefficient; Based on the feature weight, the first similarity and the second similarity, the model adjustment parameter is determined. It is expressed by the following formula: in, represents the first similarity, represents the second similarity, is the natural base, is the natural logarithm, It represents the product of the norm of the data feature, the norm of the first standard feature, and the first coefficient of variation. The product of the norm of the data feature, the norm of the second standard feature, and the third coefficient of variation, Represents feature weight.

4. The scheduling optimization method based on distributed measurement and control according to claim 1 or 3, characterized in that: The first similarity is determined by the product of the norm of the data feature, the norm of the first standard feature, and the first difference coefficient; The second similarity is determined by the product of the norm of the data feature, the norm of the second standard feature, and the third difference coefficient.

5. The scheduling optimization method based on distributed measurement and control according to claim 1, characterized in that: The method further comprises: determining a scaling factor based on the first difference coefficient; The step of reducing the second difference coefficient to obtain a third difference coefficient includes: The third difference coefficient is obtained by multiplying the second difference coefficient by the scaling ratio.

6. The scheduling optimization method based on distributed measurement and control according to claim 5, characterized in that: The first difference coefficient is used to characterize the degree of deviation between the data feature and the first standard feature, wherein the smaller the degree of deviation is, the smaller the scaling ratio is.

7. A dispatching optimization system based on distributed measurement and control, characterized in that: The method comprises a distributed measurement and control node and a cloud connected to the distributed measurement and control node in communication, wherein the distributed measurement and control node is used to detect an object to be scheduled to generate real-time node data and transmit the real-time node data to the cloud, and the cloud is configured as follows: Inputting the real-time node data into a pre-trained anomaly detection model for anomaly detection, so as to schedule the object to be scheduled according to the anomaly detection result; The training process of the anomaly detection model includes: respectively determining a first data type and a first standard feature corresponding thereto, and a second data type and a second standard feature corresponding thereto, wherein the first data type is different from the second data type; Extracting target node data from sample data according to the first data type, and extracting data features of the target node data through the model to be trained; Determine a first difference coefficient between the data feature and the first standard feature, and a second difference coefficient between the data feature and the second standard feature, and reduce the second difference coefficient to obtain a third difference coefficient; Determining a model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient, and the third difference coefficient; Train the model to be trained according to the model adjustment parameters to obtain the anomaly detection model; Wherein, determining the model adjustment parameter based on the data feature, the first standard feature, the second standard feature, the first difference coefficient and the third difference coefficient includes: Determining a first similarity based on the data feature, the first standard feature, and the first difference coefficient; Determining a second similarity based on the data feature, the second standard feature, and the third difference coefficient; Determining the model adjustment parameter based on the first similarity and the second similarity; The step of determining the model adjustment parameter based on the first similarity and the second similarity includes: Calculating a sum of the first similarity and the second similarity; Based on the ratio between the first similarity and the sum value, the model adjustment parameter is determined. It is expressed by the following formula: in, represents the first similarity, represents the second similarity, is the natural base, is the natural logarithm, It represents the product of the norm of the data feature, the norm of the first standard feature, and the first coefficient of variation. It represents the product of the norm of the data feature, the norm of the second standard feature, and the third coefficient of variation; The objects to be dispatched include various electric power equipment in the power grid to be dispatched.

Images