State monitoring method and system for alternating current metal ring network switch of power distribution system

Abstract

The invention discloses a state monitoring method and system for an AC metal ring network switch of a power distribution system, and the method comprises the steps: obtaining state data corresponding to a to-be-monitored AC metal ring network switch, determining a plurality of state types corresponding to the state data, and enabling the state data to be collected through a sensor; based on the state data, determining multiple pieces of category feature information in one-to-one correspondence with the multiple state categories; based on the state data, identifying semantic information corresponding to each category of feature information, and determining association information among a plurality of state categories; performing fusion processing on the basis of the associated information, the semantic information and the state data to obtain fused data; based on the associated information and the fusion data, an artificial intelligence model is called to determine a state monitoring result corresponding to the alternating current metal ring network switch, and the artificial intelligence model is at least constructed by a neural network, so that the state monitoring accuracy of the alternating current metal ring network switch can be improved, and the safety performance of a power distribution system is improved.

Description

Technical Field

[0001] This application relates to the field of equipment monitoring technology, and in particular to a method and system for monitoring the status of AC metal ring network switches in a power distribution system. Background Technology

[0002] AC metal ring main units are generally used in power distribution systems (such as industrial and mining enterprises, residential communities, high-rise buildings, schools, parks, etc.). They can be used as ring main power supply units (such as 12kV, 50Hz ring main power supply units) to achieve flexible power supply, rapid fault isolation and recovery.

[0003] However, when the AC metal ring main switch itself malfunctions and cannot work properly, it can easily affect the power supply reliability and fault isolation capability of the power distribution system it is located in, and may even lead to problems such as the expansion of accidents, long-term power outages for users, and equipment damage.

[0004] Therefore, accurately monitoring the status of AC metal ring network switches is crucial for improving the safety performance of power distribution systems. Summary of the Invention

[0005] To address the aforementioned technical problems, this application proposes a method and system for monitoring the status of AC metal ring network switches in a power distribution system. This method and system can improve the accuracy of status monitoring of AC metal ring network switches, thereby enhancing the safety performance of the power distribution system.

[0006] In a first aspect, embodiments of this application provide a method for monitoring the status of an AC metallic ring network switch in a power distribution system, including: Acquire status data corresponding to the AC metal ring network switch to be monitored, and determine multiple status categories corresponding to the status data, wherein the status data is acquired via sensors; Based on the state data, multiple category feature information corresponding one-to-one with the multiple state categories is determined; Based on the state data, the semantic information corresponding to each category feature information is identified, and the association information between the multiple state categories is determined; Based on the aforementioned association information, each of the aforementioned semantic information, and the aforementioned state data, a fusion process is performed to obtain fused data; Based on the associated information and the fused data, an artificial intelligence model is invoked to determine the status monitoring result corresponding to the AC metal ring network switch, wherein the artificial intelligence model is at least constructed by a neural network.

[0007] Optionally, each state category includes multiple sub-items corresponding to it, and the multiple sub-items corresponding to each state category have a hierarchical relationship in the state category; the state data includes multiple sets of data. Based on the state data, multiple category feature information corresponding one-to-one with the multiple state categories is determined, including: For each of the stated state categories, Determine the target sub-item corresponding to the state category from the sub-items corresponding to the state category, wherein the target sub-item is the lowest-level sub-item in the corresponding hierarchical relationship among the sub-items corresponding to the state category; Determine the first feature similarity between the sub-item features of the target sub-item and the data features of each of the multiple data sets; Each set of data whose first feature similarity satisfies the preset feature similarity condition is taken as the first data corresponding to the state category, and based on the first data corresponding to the state category, a category feature information corresponding to the state category is determined.

[0008] Optionally, the association information includes inter-category association information corresponding to each of the plurality of state categories; Among these, determining the association information between the multiple state categories based on the state data includes: For each of the stated state categories, Each data set other than the first data set corresponding to the state category is identified as the second data set corresponding to the state category, and the second feature similarity between the sub-item feature of the target sub-item corresponding to the state category and the data feature of each data set in the second data set is determined. Determine that there are one or more data sets in the second data where the similarity of the second feature is greater than the second feature similarity threshold; Based on the one or more data sets and their respective second feature similarities, as well as the multiple state categories, inter-category association information corresponding to the state category is generated.

[0009] Optionally, generating inter-category association information corresponding to a state category based on the one or more data sets and their respective corresponding second feature similarities, as well as the multiple state categories, includes: Determine the sub-item features of the target sub-item corresponding to each of the multiple state categories other than the current state category, and the third feature similarity between them and the data features of each of the one or more data sets. For each of the one or more datasets, if the similarity of any one or more third features corresponding to that dataset is greater than the similarity of the second feature corresponding to that dataset, then based on the state category and the state category corresponding to each of the one or more third feature similarities, inter-category association information corresponding to that state category is generated.

[0010] Optionally, the fusion processing based on the association information, each of the semantic information, and the state data to obtain fused data includes: Cross-attention calculation is performed based on each of the semantic information and the associated information to obtain the attention information corresponding to each of the semantic information. The fused data is obtained by fusing the attention information and the state data.

[0011] Optionally, the neural network includes a first expert network and a second expert network with progressively increasing complexity; The step of determining the status monitoring result corresponding to the AC metal ring network switch by calling an artificial intelligence model based on the associated information and the fused data includes: Based on the associated information and the fused data, the first expert network is invoked to generate first state analysis information; Based on the first state analysis information, the second expert network is invoked to generate second state analysis information; The status monitoring result is determined based on the second status analysis information.

[0012] Optionally, the first expert network includes P first expert modules; The step of generating first state analysis information by invoking the first expert network based on the association information and the fused data includes: Determine the association representation information corresponding to the association information and the fusion representation information corresponding to the fusion data; Based on the associated representation information and the fused representation information, the network input information is determined; Based on the network input information, t first expert modules are determined from the P first expert modules, where t≤P; Based on the network input information, the t first expert modules are invoked for processing to obtain the first state analysis information.

[0013] Optionally, the second expert network includes K second expert modules, P < K, each first expert module corresponds to at least one second expert module, and each second expert module corresponds to only one first expert module; The step of generating second state analysis information by invoking the second expert network based on the first state analysis information includes: Determine T second expert modules from among the K second expert modules that correspond to the t first expert modules, where t ≤ T; Based on the first state analysis information, s second expert modules are determined from the T second expert modules, where s≤T; Based on the first state analysis information, the s second expert modules are invoked for processing to obtain the second state analysis information.

[0014] Optionally, determining the state monitoring result based on the second state analysis information includes: From the knowledge base that matches the AC metal ring network switch, at least one piece of knowledge information corresponding to the second state analysis information is retrieved; Based on the second state analysis information and the at least one knowledge information, the initial state monitoring result is determined through the state monitoring model; Based on the second state analysis information, the at least one knowledge information and the initial state monitoring results, a knowledge density detection result related to the at least one knowledge information is generated through a knowledge density detection model. If the knowledge density detection result does not meet the knowledge density condition, then the at least one knowledge information is updated according to the knowledge density detection result, or the at least one knowledge information is updated according to the knowledge density detection result and the knowledge base, and the initial state monitoring result and subsequent steps are determined by the state monitoring model based on the second state analysis information and the at least one knowledge information. If the knowledge density detection result satisfies the knowledge density condition, the state monitoring result is determined based on the initial state monitoring result.

[0015] Secondly, embodiments of this application provide a status monitoring system for AC metallic ring network switches in a power distribution system, comprising: The data acquisition module is used to acquire status data corresponding to the AC metal ring network switch to be monitored, and to determine multiple status categories corresponding to the status data, wherein the status data is acquired by a sensor. The category feature determination module is used to determine multiple category feature information that correspond one-to-one with the multiple state categories based on the state data; The state data processing module is used to identify the semantic information corresponding to each category feature information based on the state data, and to determine the association information between the multiple state categories. The fusion module is used to perform fusion processing based on the association information, the semantic information, and the state data to obtain fused data; The status analysis module is used to determine the status monitoring result corresponding to the AC metal ring network switch by calling an artificial intelligence model based on the associated information and the fused data, wherein the artificial intelligence model is at least constructed by a neural network.

[0016] In summary, the embodiments of this application have at least the following beneficial effects: In this embodiment, status data corresponding to the AC metal ring main switch to be monitored is acquired, and multiple status categories corresponding to the status data are determined, wherein the status data is acquired by sensors; based on the status data, multiple category feature information corresponding one-to-one with the multiple status categories is determined; based on the status data, semantic information corresponding to each category feature information is identified, and correlation information between the multiple status categories is determined; based on the correlation information, each semantic information, and the status data, fusion processing is performed to obtain fused data; based on the correlation information and the fused data, an artificial intelligence model is invoked to determine the status monitoring result corresponding to the AC metal ring main switch, wherein the artificial intelligence model is at least constructed by a neural network. This improves the accuracy of status monitoring of the AC metal ring main switch, thereby enhancing the safety performance of the power distribution system. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the status monitoring method for AC metal ring network switches in a power distribution system provided in an embodiment of this application. Figure 2 This is a schematic diagram of the status monitoring system for AC metal ring network switches in a power distribution system provided in an embodiment of this application; Figure 3 This is a schematic diagram of the computer device provided in the embodiments of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments / examples are only a part of the embodiments / examples of this application, and not all of the embodiments / examples. Based on the embodiments / examples in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0019] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "multiple" means two or more. In the description of this application, the term "comprising" and its variations are open-ended, meaning "including but not limited to." The term "based on" means "at least partially based on." The term "according to" means "at least partially according to." The term "one embodiment / example" means "at least one embodiment / example"; the term "another embodiment / example" means "at least one additional embodiment / example"; the term "some embodiments / examples" means "at least some embodiments / examples."

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] 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 commonly understood by one of ordinary skill in the art. The terminology used in this application is for the purpose of describing specific embodiments only and is not intended to limit the application. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0022] Firstly, see [the following] Figure 1 The diagram shows a schematic flowchart of a method for monitoring the status of an AC metal ring main switch in a power distribution system according to an embodiment of this application. This method for monitoring the status of an AC metal ring main switch in a power distribution system can be applied to a computer device with data processing capabilities. The method includes steps S101-S105, as detailed below.

[0023] S101, acquire the status data corresponding to the AC metal ring network switch to be monitored, and determine multiple status categories corresponding to the status data, wherein the status data is acquired by a sensor.

[0024] In some examples, the sensor may include electrical sensors (such as voltage sensors, current sensors, etc.), magnetic sensors, temperature sensors, humidity sensors, gas sensors, mechanical sensors, speed / accelerometers, etc.

[0025] In some examples, at least some sensors may be installed on the AC metal ring main switch to detect the AC metal ring main switch and collect at least some status data, and / or at least some sensors may be installed in the power distribution system in which the AC metal ring main switch is located to detect the power distribution system and collect at least some status data.

[0026] In some examples, since the sensors used to acquire the state data can be predetermined, the multiple state categories corresponding to the acquired state data can be determined based on the sensor category to which each sensor belongs.

[0027] In other examples, a general large language model can be used to identify multiple state categories corresponding to the state data based on the state data and state category identification prompts. The state category identification prompts can guide the large language model to classify the state data. In this case, the identified multiple state categories can be the state categories corresponding to sub-items in the state data that the large language model initially identifies as abnormal. The large language model can be any general model commonly used in this field, and will not be specifically described here.

[0028] In some examples, the multiple state categories may include at least one of the following: electrical operating state corresponding to electrical sensors and / or magnetic sensors, mechanical action state corresponding to mechanical sensors and / or velocity / acceleration sensors, environmental adaptation state corresponding to temperature sensors, humidity sensors and / or gas sensors, insulation health state, etc.

[0029] S102, Based on the state data, determine multiple category feature information that correspond one-to-one with the multiple state categories.

[0030] In some examples, the category feature information can be generated based on at least one piece of data in the state data that matches the corresponding state category. For example, the at least one piece of data can be fused / merged, and the features of the fused / merged data can be extracted to generate the category feature information. Alternatively, feature information can be extracted from the at least one piece of data separately and fused to generate the category feature information. This is not specifically limited here.

[0031] Since each state category has some related characteristics, the related characteristics corresponding to each state category can be constructed into a corresponding category feature information. At this time, each related characteristic corresponding to each state category can correspond one-to-one with at least one set of data matching that state category. For example, the related characteristics of the insulation health state may include the insulation dielectric loss factor (tanδ), leakage current value, partial discharge amplitude and frequency and / or insulation resistance value; the related characteristics of the electrical operation state may include the three-phase voltage amplitude, three-phase current effective value, power factor, load rate and / or short-circuit current peak value; the related characteristics of the mechanical action state may include the opening and closing coil current waveform, contact stroke displacement, opening and closing time, operating mechanism vibration frequency and / or contact contact pressure; the related characteristics of the environmental adaptation state may include the cabin temperature, humidity, gas concentration and / or dust content of the AC metal ring network switch.

[0032] In some examples, multiple category features can be determined from state data using a pre-trained category feature extraction model. This model can be a trained model capable of predicting using state data as input and multiple category features as output. During training, sample state data (which also carries the expected corresponding category feature label, representing the expected category feature information) can be used to obtain the predicted category feature information generated by the model based on the sample data. Based on the difference between the predicted category feature information and the expected category feature information represented by the label, a general loss function is used to calculate the loss value. A general training algorithm (such as gradient descent) is then used to train the model based on the loss value, so that the trained model can possess the aforementioned capabilities. For example, the model may include an input representation layer, a feature fusion and representation layer, and a prediction output layer. The input representation layer can receive data from the input model and convert the data into the desired feature vector form. For example, the input representation layer can use word embeddings or pre-trained language models (such as bidirectional language representation models based on the Transformer architecture) to generate semantic vectors, and / or use embedding layers to generate dense vectors. The feature fusion and representation layer can be used to fuse the feature vectors converted by the input representation layer. For example, the feature fusion and representation layer can implement the fusion through fully connected layers or attention mechanism layers. The prediction output layer can be used to generate prediction results based on the fused features. For example, the prediction output layer can use a softmax layer to output the probability distribution for different categories, and then output the prediction result based on the probability (for example, it can output the top one or more classification results with the highest probability as the prediction result).

[0033] S103, based on the state data, identify the semantic information corresponding to each category feature information, and determine the association information between the multiple state categories.

[0034] In some examples, since the aforementioned category feature information can be used to characterize feature values ​​(or their ranges), the semantic information can be used to indicate the relevant semantics of the corresponding category feature information. For example, when the category feature information characterizes a deviation in the effective value of the three-phase current greater than a preset threshold (e.g., 15%), the corresponding semantic information can be used to indicate an imbalance in the three-phase current, thus indicating a risk of uneven load distribution. This semantic information can also be used to indicate the changing trend of the corresponding category feature information.

[0035] In some examples, the feature semantic recognition prompt, the state data, and the feature information of each category can be input into a general large language model to obtain the semantic information corresponding to each category feature information output by the large language model. The feature semantic recognition prompt can be used to guide the large language model to perform semantic recognition / semantic understanding of each category feature information using the state data as a global feature. The large language model can be any large model commonly used in the field, which will not be described in detail here. It is understood that this embodiment, by using state data as a global feature, can assist the large language model in more accurately generating semantic information corresponding to the category feature information.

[0036] In some examples, category association recognition prompts, the state data, and the multiple state categories can be input into a general large language model to obtain the association information between the multiple state categories output by the large language model. The category association recognition prompts can be used to guide the large language model to identify category associations between the multiple state categories using the state data as a global feature. The large language model can be any general large model in the field, which will not be described in detail here. It is understood that this embodiment, by using state data as a global feature, can assist the large language model in more accurately mining the associations between multiple state categories.

[0037] S104, perform fusion processing based on the association information, each of the semantic information and the state data to obtain fused data.

[0038] In some examples, since the association information can represent the degree of association between multiple state categories, the overall strength of the association between each state category and other state categories can be determined based on this degree of association (for example, the overall strength of the association can be characterized by the weighted sum of the degree of association between each state category and other state categories). Then, the fusion weight corresponding to each state category can be determined based on the overall strength of the association. Since each semantic information corresponds to a category feature information (i.e., each semantic information can correspond to a state category), the semantic information corresponding to each state category can be weighted according to the fusion weight corresponding to each state category to obtain weighted semantic information. Then, all the weighted semantic information is fused with the state data (for example, each weighted semantic information can be concatenated at the corresponding position in the state data to complete the fusion), thereby completing the fusion process and obtaining fused data.

[0039] S105, based on the associated information and the fused data, an artificial intelligence model is invoked to determine the status monitoring result corresponding to the AC metal ring network switch, wherein the artificial intelligence model is at least constructed by a neural network.

[0040] In some examples, a pre-trained artificial intelligence model can be used to determine the state monitoring result based on the correlation information and fused data. This artificial intelligence model can be a model that has been trained to have the predictive ability to take the correlation information and fused data as model input and the state monitoring result as model output. In specific training, the sample correlation information and sample fused data can be used as sample data (the sample data also carries the expected corresponding result label, which represents the corresponding expected state). The predicted state generated by the model based on the sample data is obtained. Based on the difference between the predicted state and the expected state represented by the label, a general loss function is used to calculate the loss value. Based on the loss value, a general training algorithm (such as gradient descent) is used to train the model so that the trained model can have the above-mentioned ability.

[0041] In some examples, the AI ​​model can be constructed from a neural network, which may include an input representation layer, a feature fusion and representation layer, and a prediction output layer. The input representation layer receives data from the input model and converts it into the desired feature vector form. For example, it can use word embeddings or a pre-trained language model (such as a bidirectional language representation model based on the Transformer architecture) to generate semantic vectors, and / or use embedding layers to generate dense vectors. The feature fusion and representation layer fuses the feature vectors converted by the input representation layer. For example, this fusion can be achieved through fully connected layers or attention mechanisms. The prediction output layer generates a prediction result based on the fused features. For example, it can use a softmax layer to output the probability distribution for different categories, and then output the prediction result based on this probability (e.g., it can output the top one or more categories with the highest probability as the prediction result).

[0042] In this embodiment, compared to directly monitoring the collected status data, deep fusion and intelligent correlation analysis of multi-dimensional data can be achieved to improve the accuracy of status monitoring of AC metal ring main units. Specifically, the semantic information of category feature information can be converted into corresponding semantic tags by recognizing the semantic information of category feature information, so that the artificial intelligence model can directly recognize the meaning of the equipment status behind the data, thereby reducing misjudgment. Moreover, considering that most ring main unit failures are caused by the coupling of multiple factors (such as prolonged opening and closing time due to mechanical mechanism jamming, resulting in poor contact of contacts and causing local temperature rise, ultimately leading to insulation aging), the correlation information between multiple status categories is specifically analyzed. In addition to assisting the fusion processing, it is further fed back to the artificial intelligence model to assist in processing the fused data, thereby improving the accuracy of the artificial intelligence model in predicting the status of the fused data, and ultimately improving the status monitoring accuracy of AC metal ring main units and improving the safety performance of the power distribution system.

[0043] In one optional implementation, each state category includes multiple sub-items corresponding to it, and the multiple sub-items corresponding to each state category have a hierarchical relationship within the state category; the state data includes multiple sets of data. Based on the state data, multiple category feature information corresponding one-to-one with the multiple state categories is determined, including: For each of the stated state categories, Determine the target sub-item corresponding to the state category from the sub-items corresponding to the state category, wherein the target sub-item is the lowest-level sub-item in the corresponding hierarchical relationship among the sub-items corresponding to the state category; Determine the first feature similarity between the sub-item features of the target sub-item and the data features of each of the multiple data sets; Each set of data whose first feature similarity satisfies the preset feature similarity condition is taken as the first data corresponding to the state category, and based on the first data corresponding to the state category, a category feature information corresponding to the state category is determined.

[0044] In some examples, the preset feature similarity condition may include at least one of the following: the first feature similarity is greater than a preset first feature similarity threshold, or the largest M first feature similarities. M is a positive integer, for example, M is 1.

[0045] In some examples, one of the state categories can be a mechanical action state, and its top-level sub-item can be a mechanical action sub-item. Accordingly, the next level sub-item can include a circuit breaker drive sub-item, a contact movement sub-item, and a mechanism stability sub-item. Further, the next level (bottom level) sub-item corresponding to the circuit breaker drive sub-item can include a circuit breaker coil current sub-item and / or a circuit breaker coil current sub-item. The next level (bottom level) sub-item corresponding to the contact movement sub-item can include a contact stroke displacement sub-item and / or a circuit breaker closing time sub-item. The next level (bottom level) sub-item corresponding to the mechanism stability sub-item can include an operating mechanism vibration amplitude sub-item and / or a vibration frequency sub-item.

[0046] In some examples, one of the state categories can be the environment adaptation state, and its top-level sub-item can be the environment adaptation sub-item. Accordingly, the next level sub-item can include temperature and humidity sub-items and cleanliness sub-items. Further, the next level (bottom level) sub-item corresponding to the temperature and humidity sub-item can include the cabin temperature sub-item and the cabin humidity sub-item. The next level (bottom level) sub-item corresponding to the cleanliness sub-item can include the dust concentration sub-item.

[0047] In some examples, one of the state categories can be insulation performance, and its corresponding multiple sub-items can include gas insulation state sub-items with sequentially decreasing levels, SF6 gas pressure sub-items (which can be used to characterize the SF6 gas pressure corresponding to AC metal ring network switches, and the unit can be MPa), and the corresponding target sub-item can be the SF6 gas pressure sub-item.

[0048] In some examples, one of the state categories can be thermal performance, and its corresponding multiple sub-items can include conductive loop temperature rise sub-items in descending order of hierarchy, main contact temperature sub-items (which can be used to characterize the temperature of the main contact corresponding to the AC metal ring network switch), and the corresponding target sub-item can be the main contact temperature.

[0049] It should be noted that in this embodiment, each state category can include multiple sub-items corresponding to it and having a hierarchical relationship. At least one set of data corresponding to each sub-item can generally be collected. For example, taking the gas insulation state sub-item and the SF6 gas pressure sub-item as examples, the data for the SF6 gas pressure sub-item can be directly collected by a pressure sensor, while the data for the gas insulation state sub-item can be collected by a gas insulation state detection device. This gas insulation state detection device can include a pressure sensor and a microcontroller. The microcontroller can be used to generate a gas insulation index based on the data collected by the pressure sensor included in the device, using a temperature compensation algorithm, to characterize the data for the gas insulation state sub-item. It is easy to understand that in some cases, AC metal ring main units can be equipped not only with the simplest sensors but also with devices capable of performing more complex detection functions (such as the gas insulation state detection device) to monitor the state of the AC metal ring main unit from a multi-level perspective. Here, the hierarchical level of the sub-item corresponding to the device capable of performing complex detection functions is higher than the hierarchical level of the sub-item corresponding to the sensor.

[0050] Thus, it is understandable that the lowest-level sub-item was selected in this embodiment, which can achieve more refined, quantifiable, and matchable feature matching, so as to provide a high-quality data foundation for subsequent intelligent analysis and state recognition.

[0051] Furthermore, it is understood that higher-level sub-item data may introduce noise due to potential biases or model drift in their corresponding processing algorithms. Therefore, in this embodiment, determining category feature information using the target sub-item can reduce the introduced data noise.

[0052] In one optional implementation, the association information includes inter-category association information corresponding to each of the plurality of state categories; for example, the inter-category association information can be used to characterize the degree of association between the corresponding state category and other state categories.

[0053] Among these, determining the association information between the multiple state categories based on the state data includes: For each of the stated state categories, Each data set other than the first data set corresponding to the state category is identified as the second data set corresponding to the state category, and the second feature similarity between the sub-item feature of the target sub-item corresponding to the state category and the data feature of each data set in the second data set is determined. Determine that there are one or more data sets in the second data where the similarity of the second feature is greater than the second feature similarity threshold; Based on the one or more data sets and their respective second feature similarities, as well as the multiple state categories, inter-category association information corresponding to the state category is generated.

[0054] In one optional implementation, generating inter-category association information corresponding to a state category based on the one or more data sets and their respective corresponding second feature similarities, as well as the multiple state categories, includes: Determine the sub-item features of the target sub-item corresponding to each of the multiple state categories other than the current state category, and the third feature similarity between them and the data features of each of the one or more data sets. For each of the one or more datasets, if the similarity of any one or more third features corresponding to that dataset is greater than the similarity of the second feature corresponding to that dataset, then based on the state category and the state category corresponding to each of the one or more third feature similarities, inter-category association information corresponding to that state category is generated.

[0055] In this embodiment, second and third feature similarity are introduced. By comparing the similarity between the target sub-item of the current category and the second data, as well as the similarity between the target sub-item of other categories and the second data, the hidden association between state categories can be mined. Thus, the association information can be used to reflect the multi-factor coupling characteristics of the fault.

[0056] In this embodiment, the second data refers to the data in the first data that is not classified into the corresponding state category among multiple data sets of state data. However, if there is one or more data sets in the second data set with a second feature similarity greater than the second feature similarity threshold, it indicates that the second data set still has a high correlation with the state category. Therefore, the correlation between the one or more data sets and other state categories (represented by the third feature similarity) can be further considered. If the correlation between the data sets and other state categories is higher than the correlation between the data sets and the state category, it indicates that there is a high correlation between the two state categories, making both highly correlated with the same data set. Therefore, the two state categories can be associated to facilitate more accurate analysis of the state data in the future, and to explore the degree of cross-state category correlation between the two state categories as much as possible during the analysis.

[0057] It should be noted that the "similarity" described in any one or more embodiments of this application can be calculated using at least one of the following: cosine similarity, Euclidean distance, Manhattan distance, Pearson correlation coefficient, etc. It should be understood that the implementation of this similarity method is merely illustrative and not intended to limit the scope of this application.

[0058] In one optional implementation, the fusion processing based on the association information, each of the semantic information, and the state data to obtain fused data includes: Cross-attention calculation is performed based on each of the semantic information and the associated information to obtain the attention information corresponding to each of the semantic information. The fused data is obtained by fusing the attention information and the state data.

[0059] In some examples, cross-attention computation can be used to enable the related information and the semantic information to interact with each other to form corresponding attention information, which can then be fused into the state data, so that the fused data can carry the attention information on the basis of the state data.

[0060] In some examples, each attention information can be concatenated to the position in the state data corresponding to the corresponding semantic information to complete the fusion process.

[0061] In one alternative implementation, the neural network includes a first expert network and a second expert network with progressively increasing complexity. The step of determining the status monitoring result corresponding to the AC metal ring network switch by calling an artificial intelligence model based on the associated information and the fused data includes: Based on the associated information and the fused data, the first expert network is invoked to generate first state analysis information; Based on the first state analysis information, the second expert network is invoked to generate second state analysis information; The status monitoring result is determined based on the second status analysis information.

[0062] In some examples, the second state analysis information can be directly used as the state monitoring result.

[0063] In one alternative implementation, the first expert network includes P first expert modules; The step of generating first state analysis information by invoking the first expert network based on the association information and the fused data includes: Determine the association representation information corresponding to the association information and the fusion representation information corresponding to the fusion data; Based on the associated representation information and the fused representation information, the network input information is determined; Based on the network input information, t first expert modules are determined from the P first expert modules, where t≤P; Based on the network input information, the t first expert modules are invoked for processing to obtain the first state analysis information.

[0064] In some examples, a hybrid expert network may be included in at least a portion of the first expert module and / or at least a portion of the second expert module.

[0065] In some examples, the association representation information may be obtained by transforming the association information into a specific space (e.g., latent space). Similarly, the fusion representation information may also be obtained by transforming the fusion data into a specific space (e.g., latent space).

[0066] In some examples, network input information can be obtained by fusing (e.g., splicing) the associated representation information and the fused representation information; or, the fused representation information can be obtained by fusing the associated representation information and the fused representation information. In this case, self-attention calculation can be performed on the fused information to adjust the fused information to obtain network input information.

[0067] In one optional implementation, the second expert network includes K second expert modules, P < K, each first expert module corresponds to at least one second expert module, and each second expert module corresponds to only one first expert module. The step of generating second state analysis information by invoking the second expert network based on the first state analysis information includes: Determine T second expert modules from among the K second expert modules that correspond to the t first expert modules, where t ≤ T; Based on the first state analysis information, s second expert modules are determined from the T second expert modules, where s≤T; Based on the first state analysis information, the s second expert modules are invoked for processing to obtain the second state analysis information.

[0068] In some examples, the network input information can be fed into each of the P first expert modules to calculate the probability distribution information (e.g., using softmax), thereby obtaining the probability corresponding to each of the P first expert modules. Then, based on the comparison between this probability and a preset probability threshold, t first expert modules are determined from the P first expert modules. These t first expert modules can be expert modules with probabilities greater than the preset probability threshold. Here, the probability represents the degree of correlation between the network input information and the corresponding first expert module (i.e., the processing capability of the first expert module for the network input information). In this way, t first expert modules with strong processing capabilities for the network input information can be determined, thereby improving the accuracy and reliability of the processed first state analysis information.

[0069] Then, T second expert modules corresponding to t first expert modules can be determined from the K second expert modules to further process the first state analysis information. However, before processing, it is easy to understand that the above example can be used to determine s second expert modules with strong processing capabilities for the first state analysis information from the T second expert modules, so as to process the first state analysis information and improve the accuracy and reliability of the final obtained second state analysis information.

[0070] In some examples, the corresponding first expert module and second expert module can be expert networks trained with knowledge data from the same domain. At least some of the second expert modules can use a larger amount of knowledge data and / or a higher level of data annotation during training compared to their corresponding first expert modules, so that the processing power of the second expert module is stronger than that of the first expert module.

[0071] In some cases, certain related technologies can utilize a single trained large neural network, such as DNN (Deep Neural Networks), CNN (Convolutional Neural Networks), or Transformer, to construct the aforementioned artificial intelligence model. This model is then used to process correlated information and fused data to obtain the state monitoring results corresponding to the state data. However, compared to the "expert network" in the embodiments of this application, the technical solution in this case generally has the following problems.

[0072] 1. Lack of specialization: A neural network attempts to learn all types of analysis patterns, which can easily lead to a lack of specialization and poor performance on specific complex tasks. In contrast, the embodiments of this application have stronger specialization.

[0073] 2. High computational cost: Each analysis requires activating all parameters of the entire neural network, resulting in high computational overhead, especially when the model is large (e.g., a model built from the DNN, CNN, Transformer, etc.). In contrast, the embodiments of this application only activate specific related modules for computation each time, resulting in low computational cost.

[0074] 3. Poor interpretability: It is difficult to understand how the model makes decisions and it is difficult to trace which part is responsible for which analysis. In contrast, the embodiments of this application are processed by different expert modules, and each decision can be traced back to which specific module is responsible for the analysis.

[0075] 4. Insufficient scalability: When adding new features, the entire network needs to be retrained, while the embodiments of this application only require adding a new expert module and training it to add new features.

[0076] In one optional implementation, determining the state monitoring result based on the second state analysis information includes: From the knowledge base that matches the AC metal ring network switch, at least one piece of knowledge information corresponding to the second state analysis information is retrieved; Based on the second state analysis information and the at least one knowledge information, the initial state monitoring result is determined by the state monitoring model. Based on the second state analysis information, the at least one knowledge information and the initial state monitoring results, a knowledge density detection result related to the at least one knowledge information is generated through a knowledge density detection model. If the knowledge density detection result does not meet the knowledge density condition, then the at least one knowledge information is updated according to the knowledge density detection result, or the at least one knowledge information is updated according to the knowledge density detection result and the knowledge base, and the steps of determining the initial state monitoring result and subsequent steps based on the second state analysis information and the at least one knowledge information through the state monitoring model are iteratively executed until the knowledge density detection result meets the knowledge density condition. If the knowledge density detection result satisfies the knowledge density condition, the state monitoring result is determined based on the initial state monitoring result.

[0077] In some examples, retrieving at least one piece of knowledge information corresponding to the second state analysis information from a knowledge base matching the AC metal ring network switch may include: In this knowledge base, N pieces of knowledge information that match the second state analysis information are retrieved. These N pieces of knowledge information can be arranged in order, wherein the knowledge information with a higher degree of matching with the second state analysis information (e.g., represented by information similarity) is ranked higher. Calculate the ratio between the matching degree of each of the N knowledge information (excluding the first knowledge information) and the matching degree of the previous knowledge information, as the decline ratio of the matching degree of each knowledge information. If the decline ratio of the matching degree is greater than or equal to the corresponding threshold (e.g., 90% or 85%), continue to calculate the decline ratio of the matching degree of the next knowledge information. If the decline ratio of the matching degree is less than the corresponding threshold, the previous knowledge information and all the knowledge information before it constitute the at least one knowledge information.

[0078] In some examples, the state monitoring model can be a pre-trained language model. For instance, the state monitoring prompt, the second state analysis information, and the at least one knowledge information can be input into the state monitoring model to obtain an initial state monitoring result output by the model. The state monitoring prompt is suitable for instructing the state monitoring model to update the second state analysis information with the assistance of the at least one knowledge information to obtain the initial state monitoring result.

[0079] In some examples, the knowledge density detection model can be a pre-trained language model. For instance, the knowledge density detection prompt, the second state analysis information, the at least one piece of knowledge information, and the initial state monitoring result can be input into the knowledge density detection model to obtain the knowledge density detection result output by the model. The knowledge density detection prompt is suitable for instructing the model to perform detection processing on the at least one piece of knowledge information based on the second state analysis information and the initial state monitoring result to generate the knowledge density detection result. This result may include: whether the quantity of the at least one piece of knowledge information is reasonable, too much, or too little, and / or whether there is a lack of related knowledge information among the at least one piece of knowledge information.

[0080] In some examples, the knowledge density detection result may include whether the quantity of the at least one piece of knowledge information is reasonable, too much, or too little. Accordingly, the knowledge density condition may include a quantity condition, which refers to the requirement that the knowledge density detection result must contain a reasonable quantity of the at least one piece of knowledge information. Further, if the knowledge density detection result includes too much of the at least one piece of knowledge information, the quantity of the at least one piece of knowledge information can be reduced based on the knowledge density detection result (the knowledge information ranked last among the at least one piece of knowledge information is removed). If the knowledge density detection result includes too little of the at least one piece of knowledge information, new knowledge information matching the second state analysis information can be searched from the knowledge base based on the knowledge density detection result to supplement the quantity of the at least one piece of knowledge information.

[0081] In some examples, the knowledge density detection result may include whether there is a missing piece of related knowledge information between the at least one piece of knowledge information. Accordingly, the knowledge density condition may include an association condition, which requires that the knowledge density detection result must include a missing piece of related knowledge information between the at least one piece of knowledge information. Further, if the knowledge density detection result includes a missing piece of related knowledge information between the at least one piece of knowledge information, the missing related knowledge information can be queried from the knowledge base according to the missing related knowledge information indicated by the knowledge density detection result and added to the at least one piece of knowledge information.

[0082] Secondly, correspondingly, the embodiments of this application also provide a status monitoring system for AC metal ring network switches in a power distribution system, which can realize all the processes of the status monitoring method for AC metal ring network switches in a power distribution system provided in the above embodiments.

[0083] See Figure 2 This illustration shows a schematic diagram of the status monitoring system for AC metal ring main units in a power distribution system provided in an embodiment of this application. The status monitoring system 200 for AC metal ring main units in a power distribution system includes: The data acquisition module 201 is used to acquire status data corresponding to the AC metal ring network switch to be monitored, and to determine multiple status categories corresponding to the status data, wherein the status data is acquired by a sensor. The category feature determination module 202 is used to determine multiple category feature information that correspond one-to-one with the multiple state categories based on the state data; The state data processing module 203 is used to identify the semantic information corresponding to each category feature information based on the state data, and to determine the association information between the multiple state categories. The fusion module 204 is used to perform fusion processing based on the association information, the semantic information, and the state data to obtain fused data; The status analysis module 205 is used to determine the status monitoring result corresponding to the AC metal ring network switch by calling an artificial intelligence model based on the associated information and the fused data, wherein the artificial intelligence model is at least constructed by a neural network.

[0084] In one optional implementation, each state category includes multiple sub-items corresponding to it, and the multiple sub-items corresponding to each state category have a hierarchical relationship within the state category; the state data includes multiple sets of data. Based on the state data, multiple category feature information corresponding one-to-one with the multiple state categories is determined, including: For each of the stated state categories, Determine the target sub-item corresponding to the state category from the sub-items corresponding to the state category, wherein the target sub-item is the lowest-level sub-item in the corresponding hierarchical relationship among the sub-items corresponding to the state category; Determine the first feature similarity between the sub-item features of the target sub-item and the data features of each of the multiple data sets; Each set of data whose first feature similarity satisfies the preset feature similarity condition is taken as the first data corresponding to the state category, and based on the first data corresponding to the state category, a category feature information corresponding to the state category is determined.

[0085] In one optional implementation, the association information includes inter-category association information corresponding to each of the plurality of state categories; Among these, determining the association information between the multiple state categories based on the state data includes: For each of the stated state categories, Each data set other than the first data set corresponding to the state category is identified as the second data set corresponding to the state category, and the second feature similarity between the sub-item feature of the target sub-item corresponding to the state category and the data feature of each data set in the second data set is determined. Determine that there are one or more data sets in the second data where the similarity of the second feature is greater than the second feature similarity threshold; Based on the one or more data sets and their respective second feature similarities, as well as the multiple state categories, inter-category association information corresponding to the state category is generated.

[0086] In one optional implementation, generating inter-category association information corresponding to a state category based on the one or more data sets and their respective corresponding second feature similarities, as well as the multiple state categories, includes: Determine the sub-item features of the target sub-item corresponding to each of the multiple state categories other than the current state category, and the third feature similarity between them and the data features of each of the one or more data sets. For each of the one or more datasets, if the similarity of any one or more third features corresponding to that dataset is greater than the similarity of the second feature corresponding to that dataset, then based on the state category and the state category corresponding to each of the one or more third feature similarities, inter-category association information corresponding to that state category is generated.

[0087] In one optional implementation, the fusion processing based on the association information, each of the semantic information, and the state data to obtain fused data includes: Cross-attention calculation is performed based on each of the semantic information and the associated information to obtain the attention information corresponding to each of the semantic information. The fused data is obtained by fusing the attention information and the state data.

[0088] In one alternative implementation, the neural network includes a first expert network and a second expert network with progressively increasing complexity. The step of determining the status monitoring result corresponding to the AC metal ring network switch by calling an artificial intelligence model based on the associated information and the fused data includes: Based on the associated information and the fused data, the first expert network is invoked to generate first state analysis information; Based on the first state analysis information, the second expert network is invoked to generate second state analysis information; The status monitoring result is determined based on the second status analysis information.

[0089] In one alternative implementation, the first expert network includes P first expert modules; The step of generating first state analysis information by invoking the first expert network based on the association information and the fused data includes: Determine the association representation information corresponding to the association information and the fusion representation information corresponding to the fusion data; Based on the associated representation information and the fused representation information, the network input information is determined; Based on the network input information, t first expert modules are determined from the P first expert modules, where t≤P; Based on the network input information, the t first expert modules are invoked for processing to obtain the first state analysis information.

[0090] In one optional implementation, the second expert network includes K second expert modules, P < K, each first expert module corresponds to at least one second expert module, and each second expert module corresponds to only one first expert module. The step of generating second state analysis information by invoking the second expert network based on the first state analysis information includes: Determine T second expert modules from among the K second expert modules that correspond to the t first expert modules, where t ≤ T; Based on the first state analysis information, s second expert modules are determined from the T second expert modules, where s≤T; Based on the first state analysis information, the s second expert modules are invoked for processing to obtain the second state analysis information.

[0091] In one optional implementation, determining the state monitoring result based on the second state analysis information includes: From the knowledge base that matches the AC metal ring network switch, at least one piece of knowledge information corresponding to the second state analysis information is retrieved; Based on the second state analysis information and the at least one knowledge information, the initial state monitoring result is determined through the state monitoring model; Based on the second state analysis information, the at least one knowledge information and the initial state monitoring results, a knowledge density detection result related to the at least one knowledge information is generated through a knowledge density detection model. If the knowledge density detection result does not meet the knowledge density condition, then the at least one knowledge information is updated according to the knowledge density detection result, or the at least one knowledge information is updated according to the knowledge density detection result and the knowledge base, and the initial state monitoring result and subsequent steps are determined by the state monitoring model based on the second state analysis information and the at least one knowledge information. If the knowledge density detection result satisfies the knowledge density condition, the state monitoring result is determined based on the initial state monitoring result.

[0092] Thirdly, embodiments of this application provide a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the state monitoring method for AC metal ring network switches in a power distribution system as described in any of the above claims.

[0093] Fourthly, embodiments of this application provide a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the state monitoring method for AC metal ring network switches in a power distribution system as described in any of the above claims.

[0094] Fifthly, embodiments of this application provide a computer device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the steps of the state monitoring method for AC metal ring network switches of the power distribution system as described in any of the preceding claims.

[0095] See Figure 3 The computer device in this embodiment includes a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor 301, such as a status monitoring program for AC metal ring mains switches in a power distribution system. When the processor 301 executes the computer program, it implements the steps in the aforementioned embodiments of the status monitoring methods for AC metal ring mains switches in power distribution systems, for example... Figure 1 The steps S101-S105 are shown.

[0096] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 302 and executed by the processor 301 to complete this application. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the computer device.

[0097] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will understand that the schematic diagram is merely an example of a computer device and does not constitute a limitation on the computer device. It may include more or fewer components than shown, or combine certain components, or different components. For example, the computer device may also include input / output devices, network access devices, buses, etc.

[0098] The processor 301 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or the processor 301 can be any conventional processor. The processor 301 is the control center of the computer device, connecting various parts of the entire computer device through various interfaces and lines.

[0099] The memory 302 can be used to store the computer programs and / or modules. The processor 301 implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 302 and calling the data stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 302 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0100] Wherein, if the modules / units integrated into the computer device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a non-transitory computer-readable storage medium. When the computer program is executed by the processor 301, it can implement the steps of the various method embodiments described above. Wherein, the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0101] In summary, the embodiments of this application have at least the following beneficial effects: In this embodiment, status data corresponding to the AC metal ring main switch to be monitored is acquired, and multiple status categories corresponding to the status data are determined, wherein the status data is acquired by sensors; based on the status data, multiple category feature information corresponding one-to-one with the multiple status categories is determined; based on the status data, semantic information corresponding to each category feature information is identified, and correlation information between the multiple status categories is determined; based on the correlation information, each semantic information, and the status data, fusion processing is performed to obtain fused data; based on the correlation information and the fused data, an artificial intelligence model is invoked to determine the status monitoring result corresponding to the AC metal ring main switch, wherein the artificial intelligence model is at least constructed by a neural network. This improves the accuracy of status monitoring of the AC metal ring main switch, thereby enhancing the safety performance of the power distribution system.

[0102] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary hardware platforms, or it can be implemented entirely by hardware. Based on this understanding, all or part of the technical solutions of this application that contribute to the background technology can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM (Read-Only Memory) / RAM (Random Access Memory), magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0103] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications are also considered to be within the scope of protection of this application.

Claims

1. A method for monitoring the status of AC metallic ring network switches in a power distribution system, characterized in that, include: Acquire status data corresponding to the AC metal ring network switch to be monitored, and determine multiple status categories corresponding to the status data, wherein the status data is acquired via sensors; Based on the state data, multiple category feature information corresponding one-to-one with the multiple state categories is determined; Based on the state data, the semantic information corresponding to each category feature information is identified, and the association information between the multiple state categories is determined; Based on the aforementioned association information, each of the aforementioned semantic information, and the aforementioned state data, a fusion process is performed to obtain fused data; Based on the associated information and the fused data, an artificial intelligence model is invoked to determine the status monitoring result corresponding to the AC metal ring network switch, wherein the artificial intelligence model is at least constructed by a neural network.

2. The method according to claim 1, characterized in that, Each state category includes multiple sub-items corresponding to it, and the multiple sub-items corresponding to each state category have a hierarchical relationship within the state category. The state data includes multiple sets of data. Based on the state data, multiple category feature information corresponding one-to-one with the multiple state categories is determined, including: For each of the stated state categories, Determine the target sub-item corresponding to the state category from the sub-items corresponding to the state category, wherein the target sub-item is the lowest-level sub-item in the corresponding hierarchical relationship among the sub-items corresponding to the state category; Determine the first feature similarity between the sub-item features of the target sub-item and the data features of each of the multiple data sets; Each set of data whose first feature similarity satisfies the preset feature similarity condition is taken as the first data corresponding to the state category, and based on the first data corresponding to the state category, a category feature information corresponding to the state category is determined.

3. The method according to claim 2, characterized in that, The association information includes the inter-category association information corresponding to each of the multiple state categories; Among these, determining the association information between the multiple state categories based on the state data includes: For each of the stated state categories, Each data set other than the first data set corresponding to the state category is identified as the second data set corresponding to the state category, and the second feature similarity between the sub-item feature of the target sub-item corresponding to the state category and the data feature of each data set in the second data set is determined. Determine that there are one or more data sets in the second data where the similarity of the second feature is greater than the second feature similarity threshold; Based on the one or more data sets and their respective second feature similarities, as well as the multiple state categories, inter-category association information corresponding to the state category is generated.

4. The method according to claim 3, characterized in that, The step of generating inter-category association information corresponding to a state category based on the one or more data sets and their respective corresponding second feature similarities, as well as the multiple state categories, includes: Determine the sub-item features of the target sub-item corresponding to each of the multiple state categories other than the current state category, and the third feature similarity between them and the data features of each of the one or more data sets. For each of the one or more datasets, if the similarity of any one or more third features corresponding to that dataset is greater than the similarity of the second feature corresponding to that dataset, then based on the state category and the state category corresponding to each of the one or more third feature similarities, inter-category association information corresponding to that state category is generated.

5. The method according to claim 1, characterized in that, The process of fusing the associated information, the semantic information, and the state data to obtain fused data includes: Cross-attention calculation is performed based on each of the semantic information and the associated information to obtain the attention information corresponding to each of the semantic information. The fused data is obtained by fusing the attention information and the state data.

6. The method according to claim 1, characterized in that, The neural network includes a first expert network and a second expert network with progressively increasing complexity. The step of determining the status monitoring result corresponding to the AC metal ring network switch by calling an artificial intelligence model based on the associated information and the fused data includes: Based on the associated information and the fused data, the first expert network is invoked to generate first state analysis information; Based on the first state analysis information, the second expert network is invoked to generate second state analysis information; The status monitoring result is determined based on the second status analysis information.

7. The method according to claim 6, characterized in that, The first expert network comprises P first expert modules; The step of generating first state analysis information by invoking the first expert network based on the association information and the fused data includes: Determine the association representation information corresponding to the association information and the fusion representation information corresponding to the fusion data; Based on the associated representation information and the fused representation information, the network input information is determined; Based on the network input information, t first expert modules are determined from the P first expert modules, where t≤P; Based on the network input information, the t first expert modules are invoked for processing to obtain the first state analysis information.

8. The method according to claim 7, characterized in that, The second expert network includes K second expert modules, P < K, each first expert module corresponds to at least one second expert module, and each second expert module corresponds to only one first expert module; The step of generating second state analysis information by invoking the second expert network based on the first state analysis information includes: Determine T second expert modules from among the K second expert modules that correspond to the t first expert modules, where t ≤ T; Based on the first state analysis information, s second expert modules are determined from the T second expert modules, where s≤T; Based on the first state analysis information, the s second expert modules are invoked for processing to obtain the second state analysis information.

9. The method according to claim 6, characterized in that, Determining the state monitoring result based on the second state analysis information includes: From the knowledge base that matches the AC metal ring network switch, at least one piece of knowledge information corresponding to the second state analysis information is retrieved; Based on the second state analysis information and the at least one knowledge information, the initial state monitoring result is determined by the state monitoring model. Based on the second state analysis information, the at least one knowledge information and the initial state monitoring results, a knowledge density detection result related to the at least one knowledge information is generated through a knowledge density detection model. If the knowledge density detection result does not meet the knowledge density condition, then the at least one knowledge information is updated according to the knowledge density detection result, or the at least one knowledge information is updated according to the knowledge density detection result and the knowledge base, and the initial state monitoring result and subsequent steps are determined by the state monitoring model based on the second state analysis information and the at least one knowledge information. If the knowledge density detection result satisfies the knowledge density condition, the state monitoring result is determined based on the initial state monitoring result.

10. A status monitoring system for AC metallic ring network switches in a power distribution system, characterized in that, include: The data acquisition module is used to acquire status data corresponding to the AC metal ring network switch to be monitored, and to determine multiple status categories corresponding to the status data, wherein the status data is acquired by a sensor. The category feature determination module is used to determine multiple category feature information that correspond one-to-one with the multiple state categories based on the state data; The state data processing module is used to identify the semantic information corresponding to each category feature information based on the state data, and to determine the association information between the multiple state categories. The fusion module is used to perform fusion processing based on the association information, the semantic information, and the state data to obtain fused data; The status analysis module is used to determine the status monitoring result corresponding to the AC metal ring network switch by calling an artificial intelligence model based on the associated information and the fused data, wherein the artificial intelligence model is at least constructed by a neural network.

Images