A method and system for evaluating and warning the station yard status based on algorithm fusion technology

Abstract

The present invention discloses a method and system for evaluating and warning the state of a station yard based on algorithm fusion technology, which relates to the technical field of station yard monitoring and warning. It includes filtering noise and detecting outliers for large-scale real-time station yard data collected from station yard equipment according to the frequency range and noise type; using a quantum support vector machine to perform fast feature extraction and dimensionality reduction processing on the large-scale real-time station yard data that has undergone noise filtering and outlier detection to generate a low-dimensional feature vector; constructing an information-weighted graph by using the physical connections, functional dependencies, and historical fault records among station yard equipment, where each equipment is represented as a node in the graph, and the data flow relationship, energy transmission relationship, and physical connection relationship among station yard equipment are represented as edges; using the information-weighted graph to perform multiple rounds of iterative training on a graph neural network to obtain a device relationship graph neural network model; and combining the device relationship graph neural network model with the low-dimensional feature vector to generate a final device state prediction vector.

Description

Technical Field

[0001] The present invention relates to the technical field of station field monitoring and early warning, and particularly to a method and system for evaluating and early warning the state of a station field based on algorithm fusion technology. Background Art

[0002] In the existing station field state monitoring technology, traditional methods mostly rely on single algorithms or static models, which have significant limitations. First, traditional data preprocessing is vulnerable to the influence of mixed noise frequency bands, resulting in the residue of high-frequency interference or the misdeletion of low-frequency effective signals. Feature extraction mostly relies on linear dimensionality reduction methods (such as PCA) or classical support vector machines, which are difficult to capture high-dimensional non-linear feature associations, and the computational efficiency decreases significantly with the rapid increase in the amount of data, unable to meet the real-time requirements.

[0003] Traditional equipment relationship modeling usually only considers the physical connection topology, ignoring the dynamic coupling of data flow, energy transmission, and the historical fault propagation path, resulting in the deviation of equipment state evaluation from the real risk. For example, the fault correlation between a base station and a cooling system is often simplified to independent analysis, and the chain effect of abnormal heat load on the performance of communication equipment cannot be quantified, resulting in an increase in the false alarm rate of early warning. The above defects seriously restrict the accurate state evaluation and early warning timeliness of complex station fields. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for evaluating and early warning the state of a station field based on algorithm fusion technology to solve the problems that traditional data preprocessing is vulnerable to the influence of mixed noise frequency bands and traditional equipment relationship modeling leads to the deviation of equipment state evaluation from the real risk.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for evaluating and early warning the state of a station field based on algorithm fusion technology, which includes filtering noise and detecting outliers for large-scale real-time station field data collected from station field equipment according to the frequency range and noise type;

[0008] Using a quantum support vector machine to perform fast feature extraction and dimensionality reduction processing on the large-scale real-time station field data after noise filtering and outlier detection to generate low-dimensional feature vectors;

[0009] Constructing an information-weighted graph by using the physical connection, functional dependence, and historical fault records between station field equipment, where each equipment is represented as a node, and the data flow relationship, energy transmission relationship, and physical connection relationship between station field equipment are represented as edges;

[0010] Using the information-weighted graph to perform multiple rounds of iterative training on a graph neural network to obtain an equipment relationship graph neural network model;

[0011] Combine the device relationship graph neural network model with the low-dimensional feature vector to generate a final device state prediction vector and perform state evaluation;

[0012] Judge the state warning requirement based on the evaluation result.

[0013] As a preferred solution of the station field state evaluation and warning method based on the algorithm fusion technology of the present invention, wherein: the station field devices include base station antennas, transmitters, receivers, routers, switches and cooling devices;

[0014] The noise filtering is median filtering; the outlier detection is the interquartile range method.

[0015] As a preferred solution of the station field state evaluation and warning method based on the algorithm fusion technology of the present invention, wherein: the rapid feature extraction and dimensionality reduction processing are carried out, and the specific steps are as follows,

[0016] Input the real-time station field data processed by the median filtering and the interquartile range method into the quantum support vector machine, select the quantum kernel function based on the quantum phase estimation, perform non-linear mapping on the real-time station field data, and obtain the mapped high-dimensional features;

[0017] Use the quantum computing simulator to perform parallel computing on the mapped high-dimensional features, and accelerate the generation of the feature vector through the superposition state of the quantum bits;

[0018] Based on the feature weight distribution of the quantum kernel function, adopt a sparsification strategy to screen key features, and compress the feature dimension to the target dimension through principal component analysis to generate a low-dimensional feature vector.

[0019] As a preferred solution of the station field state evaluation and warning method based on the algorithm fusion technology of the present invention, wherein: the construction of the information weighted graph is carried out, and the specific steps are as follows,

[0020] Use the physical connection, functional dependence and historical fault records between the station field devices to construct a preliminary relationship graph of the station field devices, and obtain the preliminary device nodes and connection relationships;

[0021] Adopt a weighting method to weight the relationships between nodes, and obtain weighted edges according to various factors such as data flow, energy transmission and physical connection between devices;

[0022] Use the historical fault records and fault propagation path analysis to adjust the weights of the edges, and further refine the weighted information weighted graph;

[0023] Convert the device relationship graph into an information weighted graph through the relationships between the nodes and edges in the information weighted graph.

[0024] As a preferred embodiment of the method for evaluating and warning the station field state based on the algorithm fusion technology of the present invention, wherein: the graph neural network is trained iteratively for multiple rounds using the information weighted graph, and the specific steps are as follows,

[0025] Utilize the nodes and edges in the information weighted graph to initialize the graph neural network model and obtain a preliminary device relationship graph neural network structure;

[0026] Adopt the message passing mechanism of the graph neural network to perform information passing and node update in the preliminary device relationship graph neural network structure, and obtain a preliminarily trained device relationship network;

[0027] Utilize multiple rounds of iterative training, through backpropagation and gradient update, continuously optimize the weights between nodes and the structure of the graph, and obtain a device relationship graph neural network model;

[0028] Evaluate the graph neural network after multiple rounds of training. By comparing the prediction results with the actual fault data, further adjust the device relationship graph neural network model to obtain a trained device relationship graph neural network model.

[0029] As a preferred embodiment of the method for evaluating and warning the station field state based on the algorithm fusion technology of the present invention, wherein: combine the device relationship graph neural network model with the low-dimensional feature vector to generate a final device state prediction vector and perform state evaluation. The specific steps are as follows,

[0030] Utilize the device relationship graph neural network model to combine the state information of each node in the graph with the low-dimensional feature vector to obtain the comprehensive state information of the device;

[0031] Adopt the weighted average method to combine the low-dimensional feature vector with the comprehensive state information of the device to generate a device state prediction vector of the device;

[0032] Utilize the generated device state prediction vector to evaluate the health state of the device and evaluate whether the device is in a normal working state.

[0033] As a preferred embodiment of the method for evaluating and warning the station field state based on the algorithm fusion technology of the present invention, wherein: judge the state warning requirement according to the evaluation result. The specific steps are as follows,

[0034] Based on the state evaluation result, classify the device, determine whether there are potential fault hazards, and evaluate its influence range and risk level;

[0035] Finally, according to the evaluation result, judge whether it is necessary to trigger a fault warning and take corresponding emergency measures.

[0036] Second aspect, the present invention provides a station field status evaluation and early warning system based on algorithm fusion technology, including a data processing module for filtering noise and detecting outliers from the large-scale real-time station field data collected from station field equipment according to the frequency range and noise type;

[0037] A quantum dimensionality reduction module for quickly extracting features and performing dimensionality reduction processing on the large-scale real-time station field data after noise filtering and outlier detection using a quantum support vector machine to generate low-dimensional feature vectors;

[0038] A relationship modeling module for constructing an information-weighted graph using the physical connections, functional dependencies, and historical fault records between station field equipment, where each equipment is represented as a node, and the data flow relationship, energy transmission relationship, and physical connection relationship between station field equipment are represented as edges;

[0039] A graph network training module for performing multiple rounds of iterative training on a graph neural network using the information-weighted graph to obtain a device relationship graph neural network model;

[0040] A status evaluation module for combining the device relationship graph neural network model with the low-dimensional feature vectors to generate a final device status prediction vector and perform status evaluation;

[0041] An early warning decision module for judging the need for status early warning based on the evaluation results.

[0042] Third aspect, the present invention provides a computer device including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the station field status evaluation and early warning method based on algorithm fusion technology as described in the first aspect of the present invention is implemented.

[0043] Fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, and: when the computer program is executed by the processor, any step of the station field status evaluation and early warning method based on algorithm fusion technology as described in the first aspect of the present invention is implemented.

[0044] The beneficial effects of the present invention are as follows: By integrating a variety of advanced algorithm technologies, the accuracy and real-time performance of the state evaluation and early warning of station yard equipment are significantly improved. For the large-scale real-time station yard data collected from station yard equipment, median filtering and interquartile range methods are used for noise filtering and outlier detection, effectively purifying the data, reducing interference, and improving the data quality for subsequent analysis. Using quantum support vector machines, through quantum phase estimation kernel functions and quantum bit superposition state calculations, fast feature extraction and dimensionality reduction processing are realized, generating low-dimensional feature vectors, greatly improving the data processing efficiency, and providing guarantee for real-time monitoring. An information weighted graph is constructed based on the physical connections, functional dependencies, and historical failure records between equipment, and combined with multiple rounds of iterative training of graph neural networks, the relationship characteristics between equipment are fully mined, making the state evaluation more comprehensive and accurate. By comprehensively generating a state prediction vector from the low-dimensional feature vector and the equipment relationship graph neural network model, and making early warning judgments based on the evaluation results, potential faults can be detected in a timely manner, the risk level can be evaluated, and emergency measures can be triggered. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1 It is a flowchart of the station yard state evaluation and early warning method based on algorithm fusion technology in Embodiment 1.

[0047] Figure 2 It is a module diagram of the station yard state evaluation and early warning system based on algorithm fusion technology in Embodiment 1.

[0048] Figure 3 It is a schematic diagram of quantum feature acceleration in Embodiment 1.

[0049] Figure 4 It is a dynamic weight adjustment mechanism diagram in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings of the specification.

[0051] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0052] Second, the "one embodiment" or "embodiment" mentioned herein refers to specific features, structures or characteristics that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or selectively exclusive embodiments from other embodiments.

[0053] Embodiment 1, referring to Figures 1 to 4 , is the first embodiment of the present invention. This embodiment provides a method for evaluating and warning the station field state based on algorithm fusion technology, including the following steps:

[0054] S1. Filter the noise and detect outliers for the large-scale real-time station field data collected from station field equipment according to the frequency range and noise type.

[0055] Specifically,

[0056] Use the median filtering technology to process the large-scale real-time station field data collected from station field equipment (including base station antennas, transmitters, receivers, routers, switches and cooling equipment). Select an appropriate window size according to the equipment data frequency range and noise type. For example, for the high-frequency data of base station antennas, the window size is set to 5, that is, centered on each data point, the median is calculated by combining the 2 data points before and after (a total of 5 points) to smooth the high-frequency noise; for the low-frequency data of cooling equipment, the window size is also 5.

[0057] Use the interquartile range (IQR) method to detect outliers. The specific steps are as follows: calculate the 25% quantile (Q1) and 75% quantile (Q3) of the station field data, IQR = Q3 - Q1, and the outlier threshold is set to 1.5 times IQR. The data points below Q1 - 1.5IQR or above Q3 + 1.5IQR are regarded as outliers and removed.

[0058] Furthermore, in view of the multi-source heterogeneous characteristics of the station field equipment data, a median filtering algorithm with an adaptive window size is designed. Dynamically adjust the window size according to the equipment type and data acquisition frequency. For example, for the high-frequency data of base station antennas, the window size is reduced to 3 to retain more details; for the low-frequency data of cooling equipment, the window is increased to 7 to enhance the noise suppression effect. Achieve a balance between noise suppression and data fidelity through adaptive adjustment.

[0059] It should be noted that the selection of the median filtering and IQR methods is based on the time series characteristics and noise distribution characteristics of the station field equipment data. The median filtering can effectively remove high-frequency noise without introducing too much smoothing distortion, and the IQR method is suitable for detecting outliers in non-normal distribution data. Provide a high-quality data basis for subsequent feature extraction.

[0060] S2. Use a quantum support vector machine to perform fast feature extraction and dimensionality reduction on the large-scale real-time station data that has undergone noise filtering and outlier detection, generating low-dimensional feature vectors.

[0061] Specifically, input the real-time station data processed by median filtering and the IQR method into a quantum support vector machine (QSVM). Select a quantum kernel function based on quantum phase estimation to perform a non-linear mapping on the processed real-time station data, generating a high-dimensional feature space. The number of qubits is 4, the feature mapping uses ZZFeatureMap(), and the mapping depth is 2. Use a quantum computing simulator (such as Qiskit) to perform parallel computing on the mapped high-dimensional features, accelerating the generation of feature vectors through the superposition state of qubits. Based on the feature weight distribution of the quantum kernel function, adopt a sparsification strategy to screen key features, and then compress the feature dimension to the target dimension, such as reducing it to 10 dimensions, generating low-dimensional feature vectors.

[0062] Furthermore, a quantum-accelerated feature selection and dimensionality reduction method is proposed. Utilize the parallelism of quantum computing to quickly calculate the correlations between low-dimensional feature vectors, and accelerate the eigenvalue decomposition process of PCA through the superposition and interference characteristics of quantum states. Significantly improve the computing efficiency when processing large-scale and high-dimensional station equipment data.

[0063] It should be noted that QSVM combines the advantages of quantum computing and can efficiently process high-dimensional data, especially suitable for scenarios where station equipment data is large in volume and high in dimension. Through fast feature extraction and dimensionality reduction, not only is the computational complexity reduced, but also key information is retained, laying a foundation for subsequent training.

[0064] S3. Construct an information-weighted graph using the physical connections, functional dependencies, and historical fault records between station equipment. Each equipment is represented as a node, and the data flow relationship, energy transmission relationship, and physical connection relationship between station equipment are represented as edges.

[0065] Specifically, based on the physical connections (such as cables and optical fibers), functional dependencies (such as data streams and control signals), and historical fault records between station equipment, construct preliminary equipment nodes and connection relationships. Each equipment serves as a node, the physical connection is an undirected edge, and the functional dependency is a directed edge. Adopt a weighting method to calculate the initial weights of the edges according to the data flow intensity, energy transmission volume, and physical connection reliability between equipment. Data flow weight = data transmission rate / maximum rate. Energy transmission weight = energy transmission volume / maximum energy. Physical connection weight = connection reliability coefficient (range 0 - 1).

[0066] Furthermore, a method for constructing a dynamic weighted graph based on multi-source information is proposed. By combining physical connections, functional dependencies, and historical fault data, the weights of the graph are updated in real time, enabling the graph structure to dynamically reflect the real relationships and potential risks among devices.

[0067] It should be noted that the information weighted graph is not only a representation of the static connections of devices, but also incorporates dynamic data flows and historical fault information. This comprehensive design enables the information weighted graph to accurately reflect the operating status of devices and potential fault propagation paths, providing a reliable basis for subsequent analysis.

[0068] S4. Use the information weighted graph to perform multiple rounds of iterative training on the graph neural network to obtain a device relationship graph neural network model.

[0069] Specifically, use the information weighted graph to initialize the graph neural network (GNN), adopting the graph convolutional network (GCN) structure. The number of layers is 3, the hidden layer dimension is 64, the activation function is ReLU. The learning rate is 0.01, the number of training rounds is 100, and the loss function is mean squared error (MSE). Use actual fault data to verify the model performance and adjust the hyperparameters to improve the accuracy.

[0070] Furthermore, dynamically adjust the learning rate according to the change of the loss during the training process. For example, when the loss decreases slowly, reduce the learning rate to avoid overfitting or underfitting, thereby improving the model convergence speed and prediction accuracy.

[0071] It should be noted that through multiple rounds of iterative training, the complex relationships among devices can be effectively captured. The learned model not only reflects the static connections but also reveals the potential associations among dynamic states, providing strong support for subsequent state prediction.

[0072] S5. Combine the device relationship graph neural network model with the low-dimensional feature vector to generate the final device state prediction vector and perform state evaluation.

[0073] Specifically, combine the node representation output by the device relationship graph neural network model with the low-dimensional feature vector, and adopt the feature concatenation technique to generate a comprehensive feature vector. Use the random forest method to combine the comprehensive feature vector to generate the device state prediction vector. The number of trees is 100, and the maximum depth is 10. Evaluate the device state according to the prediction vector and set the evaluation threshold: a predicted value < 0.3 is normal, 0.3 - 0.7 is a warning, and > 0.7 is a fault.

[0074] Furthermore, by combining the graph structure information of GNN and the classification ability of random forest, the robustness and accuracy of the prediction are improved, and the limitations of a single model are reduced.

[0075] It should be noted that by integrating GNN and low-dimensional feature vectors, the relationships between devices and individual characteristics are comprehensively considered, making the state assessment more comprehensive and reliable, and providing a solid foundation for early warning judgment.

[0076] S6. Judge the state early warning requirements based on the evaluation results.

[0077] Specifically, according to the evaluation results of S5, the devices are divided into three categories: normal, potential failure, and high-risk failure. The potential failure corresponds to a predicted value in the range of 0.3 to 0.7, and the high-risk failure corresponds to a predicted value > 0.7. Use the information-weighted graph to analyze the influence range and risk level of potential failure devices, identify the fault propagation path, and evaluate the number and criticality of affected devices. Decide whether to trigger an early warning according to the risk level. The risk level is divided into low, medium, and high, and medium and high risks trigger an early warning.

[0078] Furthermore, by simulating the fault propagation process, calculate the risk index of each device to provide quantitative support for early warning decision-making.

[0079] It should be noted that the early warning requirement judgment combines state assessment and risk analysis, can timely detect potential failures and evaluate the influence range, thus providing a decision-making basis for emergency measures and ensuring the stable operation of station yard equipment.

[0080] This embodiment also provides a station yard state assessment and early warning system based on algorithm fusion technology, including:

[0081] A data processing module, which is used to perform noise filtering and outlier detection on the large-scale real-time station yard data collected from station yard equipment according to the frequency range and noise type;

[0082] A quantum dimensionality reduction module, which is used to perform fast feature extraction and dimensionality reduction processing on the large-scale real-time station yard data that has undergone noise filtering and outlier detection using quantum support vector machines to generate low-dimensional feature vectors;

[0083] A relationship modeling module, which is used to construct an information-weighted graph using the physical connections, functional dependencies, and historical fault records between station yard equipment. Each device is represented as a node, and the data flow relationship, energy transmission relationship, and physical connection relationship between station yard equipment are represented as edges;

[0084] A graph network training module, which is used to perform multiple rounds of iterative training on the graph neural network using the information-weighted graph to obtain a device relationship graph neural network model;

[0085] A state assessment module, which is used to combine the device relationship graph neural network model with the low-dimensional feature vectors to generate a final device state prediction vector and perform state assessment;

[0086] An early warning decision-making module, which is used to judge the state early warning requirements based on the evaluation results.

[0087] This embodiment also provides a computer device, which is applicable to the situation of the station field state evaluation and early warning method based on the algorithm fusion technology, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the station field state evaluation and early warning method based on the algorithm fusion technology proposed in the above embodiment.

[0088] This computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad set on the outer shell of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0089] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the station field state evaluation and early warning method based on the algorithm fusion technology proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (abbreviated as SRAM), electrically erasable programmable read-only memory (abbreviated as EEPROM), erasable programmable read-only memory (abbreviated as EPROM), programmable read-only memory (abbreviated as PROM), read-only memory (abbreviated as ROM), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0090] In summary, the present invention: by integrating a variety of advanced algorithm technologies, significantly improves the accuracy and real-time performance of the state evaluation and early warning of station equipment. For the large-scale real-time station data collected from station equipment, the median filtering and interquartile range methods are used for noise filtering and outlier detection, effectively purifying the data, reducing interference, and improving the data quality for subsequent analysis. Using quantum support vector machines, through quantum phase estimation kernel functions and quantum bit superposition state calculations, fast feature extraction and dimensionality reduction processing are realized, generating low-dimensional feature vectors, greatly improving the data processing efficiency, and providing guarantee for real-time monitoring. An information weighted graph is constructed based on the physical connections, functional dependencies, and historical fault records between equipment, and combined with multiple rounds of iterative training of graph neural networks, fully mining the relationship characteristics between equipment, making the state evaluation more comprehensive and accurate. By comprehensively generating a state prediction vector from the low-dimensional feature vector and the equipment relationship graph neural network model, and making an early warning judgment based on the evaluation results, potential faults can be discovered in a timely manner, the risk level can be evaluated, and emergency measures can be triggered.

[0091] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for evaluating and warning the station yard status based on algorithm fusion technology, characterized in that: including, filtering noise and detecting outliers for large-scale real-time station data collected from station equipment according to the frequency range and noise type; using a quantum support vector machine to extract features and reduce the dimension of the large-scale real-time station data after noise filtering and outlier detection, generating a low-dimensional feature vector; constructing an information-weighted graph using the physical connections, functional dependencies, and historical fault records between station equipment, where each equipment is represented as a node, and the data flow relationship, energy transmission relationship, and physical connection relationship between station equipment are represented as edges; using the information-weighted graph to perform multiple rounds of iterative training on the graph neural network to obtain a device relationship graph neural network model. The specific steps are as follows: initializing the graph neural network model using the nodes and edges in the information-weighted graph to obtain a preliminary device relationship graph neural network structure; adopting the message passing mechanism of the graph neural network to perform information passing and node update in the preliminary device relationship graph neural network structure to obtain a preliminarily trained device relationship network; using multiple rounds of iterative training, through backpropagation and gradient update, continuously optimizing the weights between nodes and the structure of the information-weighted graph to obtain a device relationship graph neural network model; evaluating the graph neural network after multiple rounds of training, further adjusting the device relationship graph neural network model by comparing the prediction results with the actual fault data to obtain a trained device relationship graph neural network model; combining the device relationship graph neural network model with the low-dimensional feature vector to generate a final device state prediction vector and perform state evaluation; judging the state warning requirement based on the evaluation result.

2. The method for evaluating and warning the station field state based on the algorithm fusion technology according to claim 1, characterized in that: The station equipment includes base station antennas, transmitters, receivers, routers, switches, and cooling equipment; The noise filtering is median filtering; the outlier detection is the interquartile range method.

3. The method for evaluating and warning the station yard status based on the algorithm fusion technology according to claim 2, wherein: For the feature extraction and dimension reduction processing, the specific steps are as follows: inputting the real-time station data processed by median filtering and the interquartile range method into a quantum support vector machine, selecting a quantum kernel function based on quantum phase estimation to perform a non-linear mapping on the real-time station data to obtain the mapped high-dimensional features; using a quantum computing simulator to perform parallel computing on the mapped high-dimensional features to accelerate the generation of feature vectors through the superposition state of quantum bits; based on the feature weight distribution of the quantum kernel function, adopting a sparsification strategy to screen key features and compressing the feature dimension to the target dimension through principal component analysis to generate a low-dimensional feature vector.

4. The method for evaluating and warning the station yard state based on the algorithm fusion technology according to claim 3, characterized in that: For constructing the information-weighted graph, the specific steps are as follows: using the physical connections, functional dependencies, and historical fault records between station equipment to construct a preliminary relationship graph of station equipment to obtain preliminary device nodes and connection relationships; adopting a weighting method to weight the relationships between nodes, and obtaining weighted edges based on the data flow, energy transmission, and physical connections between equipment; adjusting the weights of the edges using historical fault records and fault propagation path analysis; transforming the preliminary relationship graph into an information-weighted graph through the relationships between nodes and edges in the preliminary relationship graph.

5. The method for evaluating and warning the station field state based on the algorithm fusion technology according to claim 4, wherein: Combining the device relationship graph neural network model with the low-dimensional feature vector to generate a final device state prediction vector and perform state evaluation. The specific steps are as follows: Using the device relationship graph neural network model, combine the status information of each node in the information weighted graph with the low-dimensional feature vector to obtain the comprehensive status information of the device; Adopt the weighted average method, combine the low-dimensional feature vector with the comprehensive status information of the device to generate the device status prediction vector of the device; Use the generated device status prediction vector to evaluate the health status of the device and evaluate whether the device is in a normal working state.

6. The method for evaluating and warning the station yard state based on the algorithm fusion technology according to claim 5, wherein: Judge the status warning requirement according to the evaluation result. The specific steps are as follows. Based on the status evaluation result, classify the device, determine whether there are potential fault hazards, and evaluate the influence range and risk level; According to the evaluation result, judge whether it is necessary to trigger a fault warning and take corresponding emergency measures.

7. A station field status evaluation and early warning system based on algorithm fusion technology, based on the station field status evaluation and early warning method based on algorithm fusion technology according to any one of claims 1 to 6, characterized in that: Including A data processing module, configured to perform noise filtering and outlier detection on the large-scale real-time station data collected from the substation equipment according to the frequency range and noise type; A quantum dimensionality reduction module, configured to perform fast feature extraction and dimensionality reduction processing on the large-scale real-time station data that has undergone noise filtering and outlier detection using a quantum support vector machine to generate a low-dimensional feature vector; A relationship modeling module, configured to construct an information weighted graph using the physical connection, functional dependence, and historical fault records between the substation equipment. Each device is represented as a node, and the data flow relationship, energy transmission relationship, and physical connection relationship between the substation equipment are represented as edges; A graph network training module, configured to perform multiple rounds of iterative training on the graph neural network using the information weighted graph to obtain the device relationship graph neural network model; A status evaluation module, configured to combine the device relationship graph neural network model with the low-dimensional feature vector to generate the final device status prediction vector and perform status evaluation; A warning decision module, configured to judge the status warning requirement according to the evaluation result.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, the steps of the substation status evaluation and warning method based on the algorithm fusion technology according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of the substation status evaluation and warning method based on the algorithm fusion technology according to any one of claims 1 to 6 are implemented.

Images