Intelligent monitoring and fault prediction method and system for switch

Abstract

The invention relates to an intelligent monitoring and fault prediction method and system for a switch. The method comprises the following steps: extracting basic electric signals generated by each component in a switch on a conduction characteristic level by combining a feature decoding and state mapping combined mode to obtain a monitoring index set; according to the drift mode of each monitoring index in the monitoring index set, performing fusion evolution processing on the monitoring index set to obtain a trend label set representing an abnormal evolution trend; and according to the trend label set, evolution state analysis combining dual constraints of an abnormal evolution trajectory and a trend label is carried out on the monitoring index set, and a prediction result set used for predicting switch component faults is obtained. By adopting the method, refined monitoring and precise fault early warning of each component of the switch can be realized.

Description

Technical Field

[0001] The present application relates to the field of switches, and in particular to a method and system for intelligent monitoring and fault prediction of switches. Background Art

[0002] In the switch field, monitoring switch data is crucial for predicting component failures. Related switch monitoring and detection methods assess component status by periodically sampling and generating alarms based on thresholds for single physical parameters such as voltage, current, and temperature. However, this approach suffers from shortcomings such as reliance on a single indicator, delayed response, and difficulty characterizing the evolution of the indicator. Consequently, potential failure trends cannot be identified promptly in actual operation, and false alarms or missed alarms are common. Summary of the Invention

[0003] Based on this, it is necessary to provide a method, system, computer device and computer-readable storage medium for intelligent monitoring and fault prediction of switches to address the above technical problems, so as to achieve refined monitoring and accurate fault warning of various components of the switch.

[0004] In a first aspect, the present application provides a method for intelligent monitoring and fault prediction of a switch, comprising: By combining feature decoding with state mapping, the basic electrical signals generated by each component in the switch are analyzed at the level of conduction characteristics to obtain a set of monitoring indicators. According to the drift pattern of each monitoring indicator in the monitoring indicator set, the monitoring indicator set is subjected to fusion evolution processing to obtain a trend label set representing an abnormal evolution trend; According to the trend tag set, the monitoring indicator set is subjected to an evolution state analysis combining dual constraints of abnormal evolution trajectory and trend tag, to obtain a prediction result set for predicting switch component failure.

[0005] In a second aspect, the present application also provides an intelligent monitoring and fault prediction system for a switch, comprising: The monitoring module is used to analyze the basic electrical signals generated by each component in the switch at the level of conductivity characteristics by combining feature decoding and state mapping to obtain a set of monitoring indicators; A trend analysis module, configured to perform fusion evolution processing on the monitoring indicator set according to the drift pattern of each monitoring indicator in the monitoring indicator set, and generate a trend label set representing an abnormal evolution trajectory; The prediction module is used to perform an evolution state analysis of the monitoring indicator set based on the trend label set, combining the dual constraints of abnormal evolution trajectory and trend label, to obtain a prediction result set for predicting switch component failure.

[0006] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above steps when executing the computer program.

[0007] In a fourth aspect, the present application further provides a computer-readable storage medium on which a computer program is stored, and the computer program implements the above steps when executed by a processor.

[0008] The intelligent monitoring and fault prediction method, system, computer device and computer-readable storage medium of the above-mentioned switch firstly combines the joint method of feature decoding and state mapping to analyze the basic electrical signal at the conduction characteristic level to obtain a set of monitoring indicators, thereby realizing a structured quantitative expression of the conduction characteristic changes of each component; secondly, according to the drift pattern of each monitoring indicator, the monitoring indicator set is subjected to fusion evolution processing to construct a trend label set, thereby realizing the temporal trend classification and identification of the abnormal evolution process; thirdly, according to the trend label set, the monitoring indicator set is subjected to dual-constraint analysis processing, thereby realizing targeted prediction output of component failure risk; based on this, a complete analysis path from signal perception to trend identification to fault prediction can be systematically constructed in a data-driven manner, thereby realizing refined monitoring of each component of the switch and precise fault warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0010] Figure 1 1 is a flow chart of a method for intelligent monitoring and fault prediction of a switch in one embodiment; Figure 2 FIG. 4 is a structural block diagram of an intelligent monitoring and fault prediction system for a switch in one embodiment. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0012] In one embodiment, Figure 1As shown, a method for intelligent monitoring and fault prediction of a switch is provided. This embodiment uses the method applied to a server as an example. It is understood that the method can also be applied to a terminal, or to a system including a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps S101 to S103.

[0013] In step S101 , the basic electrical signals generated by each component in the switch are analyzed at the level of conduction characteristics by combining feature decoding with state mapping to obtain a set of monitoring indicators.

[0014] Among them, a switch refers to a device used in a network system to complete data exchange and routing control between different network nodes. It is used to implement functions such as data forwarding, protocol conversion, and path selection in a local area network or wide area network, such as an Ethernet switch deployed in an enterprise network or data center.

[0015] Among them, each component in the switch represents the hardware subsystem or circuit unit that constitutes the overall functional module of the switch, which is used to undertake specific functions such as power management, signal forwarding, protocol parsing, storage caching, etc., such as power modules, main control units, interface modules, backplane bus circuits, etc.

[0016] Among them, the basic electrical signal represents the original electrical waveform or voltage / current signal generated by each component in the switch during operation, which is used to reflect physical characteristics such as the component conduction state, load changes and communication response. For example, an interface module generates a high-frequency pulse signal or voltage stability fluctuation signal when forwarding data.

[0017] Among them, the monitoring indicator set represents a multi-dimensional parameter set used to reflect the changes in the conduction characteristics during the operation of the component, and is used to monitor the operating status of each component. For example, the conduction voltage fluctuation amplitude, response delay change trend, current disturbance frequency, etc. are parameter sets composed of monitoring indicators that represent different conduction characteristic changes.

[0018] For example, first, the original basic electrical signals are obtained from various components of the switch, such as the power module, main control unit, interface module, and backplane bus circuit. The working status of the basic electrical signals generated by these components is expressed in the form of continuous or discrete electrical signals. Furthermore, after signal collection, the basic electrical signals are analyzed at the level of conduction characteristics using a combination of feature decoding and state mapping. Specifically, the feature decoding process quantitatively decomposes the basic electrical signals into signal features in multiple dimensions, including frequency, amplitude, duration, rise time, and stable level, thereby clarifying the changes in the conduction characteristics of each component under different operating conditions. The state mapping process maps these decoded signal features against a predetermined component operation model to determine the physical or logical working state corresponding to the signal feature, and labels the corresponding state category and corresponding monitoring indicators. Based on this, through the above-mentioned combined processing method, monitoring indicators representing the changes in the conduction characteristics of each component are extracted from the basic electrical signals of each component, thereby constructing a monitoring indicator set.

[0019] Step S102 : performing fusion evolution processing on the monitoring indicator set according to the drift pattern of each monitoring indicator in the monitoring indicator set to obtain a trend label set representing an abnormal evolution trend.

[0020] Among them, the drift pattern represents a structured expression of the continuous change pattern or offset behavior of the monitoring indicators in the time dimension, which is used to identify the performance degradation or abnormal evolution trend of the components during long-term operation. For example, a monitoring indicator related to a certain current characteristic shows a change pattern of gradually increasing and exceeding the set stability threshold range.

[0021] Among them, the trend label set represents the structured classification results used to identify various types of abnormal evolution trends, that is, it is used to classify and manage the abnormal evolution behavior of the component operating status. For example, the label set is composed of trend labels representing different abnormal evolution trends, such as chronic conduction performance degradation, periodic voltage mutation, and enhanced response delay drift.

[0022] For example, first, the monitoring indicator set is subjected to time-series structured processing, that is, the values ​​of each type of monitoring indicator are organized into corresponding time series according to the sampling time, to ensure that the change process of the monitoring indicator can be accurately reflected in the time dimension. Subsequently, the change pattern of each monitoring indicator during the operation cycle is identified, that is, whether its value shows characteristic behaviors such as continuous offset, mutation, and periodic fluctuation, thereby summarizing the drift pattern of the monitoring indicator. Furthermore, for the characteristic behavior of these monitoring indicators, the correlation between multiple monitoring indicators is included in the modeling scope through multi-dimensional fusion modeling; for example, the increasing trend of the amplitude of a certain signal and the decreasing trend of the frequency of another signal may jointly characterize a specific structural aging phenomenon. Finally, the results of the above modeling process are structured and output in the form of labels, that is, a set of trend labels is generated. Each trend label represents a feature combination with a specific evolutionary significance and is used to characterize the existing abnormal evolutionary trends.

[0023] Step S103 : performing an evolution state analysis on the monitoring indicator set based on the trend tag set, combining the dual constraints of abnormal evolution trajectory and trend tag, to obtain a prediction result set for predicting switch component failure.

[0024] Among them, the evolutionary state represents a structured expression of the dynamic change position of a monitoring indicator at the current moment and its evolutionary stage, which is used to characterize the specific evolution of the component during the transition from normal operation to potential fault state.

[0025] The prediction result set represents a structured output dataset obtained by analyzing the dual constraints of abnormal evolution trajectories and trend labels to characterize the potential failure risks of switch components, which is used to support operations such as maintenance scheduling and risk intervention.

[0026] Exemplarily, the trend tag set and the original monitoring indicator set need to be subjected to dual constraint analysis processing to analyze the abnormal evolution state of the switch components during operation from the overall structure, and thereby form a prediction result set for predicting faults. Specifically, first, the trend tags in the trend tag set not only characterize the quantitative change characteristics of certain monitoring indicators, but also represent the combination rules and temporal relationships between these characteristics; based on this, combined with the change patterns of these trend tags, pattern retrieval and matching are performed in the monitoring indicator set, that is, to determine whether there is an abnormal evolution trajectory that completely corresponds to or is highly similar to a certain trend tag. If the match is successful, it means that the current monitoring indicator has entered a specific evolution state, and information such as the duration, change rate, and number of participating indicators of the evolution state can be further used to determine its degree of evolution.

[0027] Furthermore, based on the identification and analysis of the current evolutionary status of each monitoring indicator, the component evolutionary status that satisfies the dual constraints of abnormal evolution trajectory and trend label is further obtained, thereby constructing a set of prediction results describing the evolutionary process of each component failure; this prediction result set contains multiple fields such as the component number predicted to fail, fault type indication, possible time interval for the failure to occur, and the evolutionary stage of the current fault, providing a reliable data basis for taking maintenance and protection strategies in advance.

[0028] In the above-mentioned intelligent monitoring and fault prediction method of the switch, first, the basic electrical signal is analyzed at the conduction characteristic level by combining feature decoding and state mapping to obtain a set of monitoring indicators, thereby realizing a structured quantitative expression of the changes in the conduction characteristics of each component; secondly, according to the drift pattern of each monitoring indicator, the monitoring indicator set is fused and evolved to construct a trend label set, thereby realizing the temporal trend classification and identification of the abnormal evolution process; thirdly, the monitoring indicator set is subjected to dual-constraint analysis processing based on the trend label set, thereby realizing targeted prediction output of component failure risks; based on this, a complete analysis path from signal perception to trend identification to fault prediction can be systematically constructed in a data-driven manner, thereby realizing refined monitoring of each component of the switch and precise fault warning.

[0029] In an exemplary embodiment, the basic electrical signals generated by each component in the switch are analyzed at the level of conduction characteristics by combining feature decoding with state mapping to obtain a set of monitoring indicators, including steps S201 to S203.

[0030] Step S201 : By identifying key change trends in signal responses, feature decoding processing is performed on the basic electrical signals generated by each component to obtain a feature information set characterizing the conduction behavior of the component.

[0031] Among them, the key change trend in the signal response represents the fluctuation characteristics of the basic electrical signal in the time dimension that have identification significance, which is used to reveal the non-steady-state behavior or mutation law of the basic electrical signal during the conduction process, such as the steepness of the voltage rising edge, the short-term high-frequency disturbance in the current waveform, or the duration of the stable segment.

[0032] Among them, the characteristic information set that characterizes the conduction behavior of the component represents a multidimensional quantitative parameter set used to describe the conduction process of the component, that is, a parameter combination used to express the initial response, stability maintenance and energy consumption characteristics of the conduction process, such as rise time, peak amplitude, jitter frequency and other indicators.

[0033] For example, key segments reflecting state changes are first extracted from the continuous basic electrical signals during component operation. These key segments may include voltage rising edges, current mutation points, waveform stable intervals, or periodic fluctuation segments. Furthermore, key trend features are extracted from these key segments. This involves analyzing the basic electrical signals in their time series form, extracting inflection points, zero-crossing points, and stable windows present in each signal trajectory as candidate feature points. These features are then further screened based on properties such as their rate of change, fluctuation range, and repeatability. This identifies key trends in the signal response and ultimately forms a feature information set that characterizes the component's conduction behavior. This feature information set reflects the response characteristics of each component during the conduction process, including quantitative information in multiple dimensions, such as conduction start-up delay, response amplitude variation, and energy dissipation, rather than simple voltage or level values. Based on this, characteristic parameters with a high correlation with the conduction process are extracted from the original electrical signals to reflect the differences in signal behavior of each component under actual operating conditions.

[0034] Step S202 : By constructing a response association relationship between features and states, a state mapping process is performed on the conduction states of each component in the feature information set to obtain a state representation set representing the conduction state differences.

[0035] Among them, the state representation set that characterizes the difference in conduction states is a set of state type labels used to distinguish the conduction states of components, that is, it is used to identify what conduction state the component is currently in, such as a structured set containing labels such as "normal conduction", "conduction fluctuation" or "conduction instability".

[0036] For example, first, the representation dimensions of the conduction state need to be defined. These representation dimensions may include whether conduction occurs, the response stability during the conduction process, and the integrity of signal changes. Based on this, the feature information in the feature information set is structurally classified. Furthermore, based on the response patterns between the feature information and the conduction state, the characteristic value interval, change rate, duration, and other characteristics in the feature information are pre-associated with predefined conduction state labels to construct a set of mapping rules that can reflect the differences in conduction states. For example, if a feature information shows a continuous upward trend with low volatility over a period of time, it can be mapped to a "stable conduction" state; if a feature information suddenly changes or exhibits periodic irregularities, it can be mapped to a "fluctuating conduction" or "abnormal conduction" state. Furthermore, the matching relationships reflected in the above mapping rule set are applied to the classified feature information to generate a set of state representations representing different conduction states. This state representation set not only reflects the conduction state type of each component in the same time period, but also reflects the boundary conditions and classification characteristics between these conduction states, forming a structured basis for the classification and cognition of conduction behavior.

[0037] Step S203 , obtaining a matching conduction characteristic association structure according to the conduction state differences represented by the state representation set, and performing multi-dimensional modeling on the conduction characteristic changes of each component based on the conduction characteristic association structure to obtain a monitoring indicator set.

[0038] Among them, the conduction characteristic association structure represents a structured logical framework constructed by the transfer relationship and coordinated change pattern between the conduction states of components. It is used to characterize the distribution and evolution of the conduction characteristics of each component in the time and structural dimensions. For example, the state transition path structure formed by the frequent transition of a component state from "stable conduction" to "unstable conduction" is used.

[0039] For example, the differences between conduction states are systematically modeled based on a set of state representations, and a conduction characteristic association structure is constructed accordingly. Specifically, by analyzing the evolutionary paths between conduction states, the transition frequencies and durations between states, and the distribution patterns of different state combinations among components, a multidimensional graph characterizing the evolution of component conduction behavior is structurally constructed. This multidimensional graph is then mapped to form a logical model describing the trend of conduction characteristic changes, namely the conduction characteristic association structure. Subsequently, the conduction characteristics of each component are modeled based on this conduction characteristic association structure, using the state representation and structural relationship as modeling inputs, and outputting a multidimensional parameter vector that can quantitatively express the conduction change characteristics, thereby forming a set of monitoring indicators. In addition, the entire modeling process is not only based on the static differences of the states themselves, but also takes into account the dynamic evolutionary relationships between states, ensuring that the final generated set of monitoring indicators has logical continuity and structural separability.

[0040] In this embodiment, first, based on the key change trends in the signal response, the basic electrical signal is feature decoded, thereby extracting a feature information set that can reflect the conduction behavior of the component, laying the foundation for subsequent state identification; secondly, based on the response correlation relationship between the feature and the state, the conduction state of each component in the feature information set is state mapped, thereby converting the feature information into a state representation set with classifiable attributes; thirdly, based on the conduction state differences in the state representation set, a conduction characteristic association structure is constructed, thereby performing multi-dimensional modeling processing on the conduction characteristic changes of each component based on the conduction characteristic association structure to obtain a monitoring indicator set, based on which, a stable state identification system and a quantifiable monitoring indicator system are gradually constructed, providing a complete structure and accurate expression of basic data support for subsequent abnormal trend analysis and fault prediction.

[0041] In an exemplary embodiment, a matching conduction characteristic association structure is obtained according to the conduction state differences represented by the state representation set, and the conduction characteristic changes of each component are multi-dimensionally modeled based on the conduction characteristic association structure to obtain a monitoring indicator set, including steps S301 to S303.

[0042] Step S301 : performing difference calculation processing on each component in the state representation set according to the difference characteristics of the conduction states of each component in terms of change amplitude and change direction, and obtaining a state difference map representing the change relationship of the conduction states of the components.

[0043] Among them, the differential characteristics of the conduction states of various components in terms of change amplitude and change direction indicate the relative differences in the degree of numerical change and change trend path of different components during the conduction state change process. They are used to characterize the deviation relationship of the conduction behavior of each component in the time dimension. For example, if the conduction state of one component rises rapidly from a low level, while the other component remains stable, there are significant differences in both the change amplitude and direction.

[0044] Among them, the state difference map represents a graphical data structure formed after measuring and structuring the conduction state change relationship between multiple components. It is used to express the conduction behavior distance relationship between different components. For example, in a symmetric matrix, each element represents the difference value of the conduction state between two components.

[0045] For example, first, the conduction state trajectory of each component over a continuous time period is extracted from the state representation set, and these conduction state trajectories are standardized to make the conduction state sequences of different components comparable. Furthermore, by defining two key dimensions: the amplitude and direction of change between conduction states, the degree of numerical deviation of the conduction state sequence of each component and the trend direction of state change are measured. For example, if the conduction state of a component frequently shifts from a stable state to an unstable state, while another component maintains its original state, there is a significant difference in the amplitude of change between the two components, and the direction of change also shows inconsistency. Furthermore, by calculating the state difference values ​​of the above two dimensions between each component, a matrix map containing the conduction state differences between all components is constructed, namely the state difference map. This state difference map not only retains the quantitative change information of the conduction state but also expresses the relative relationship between the conduction behaviors of each component. Each element in the state difference map corresponds to a state difference value, which can be used to measure the impact of a component on the overall conduction characteristics and reveal potential behavioral coupling characteristics between components.

[0046] Step S302 : clustering components having correlation in conduction behaviors according to the state difference map to obtain a conduction characteristic correlation structure representing the correlation of conduction behaviors.

[0047] Among them, the conduction behavior correlation indicates the coupling degree or synchronization characteristics of the conduction state changes of multiple components during operation. It is used to identify groups of components that behave in a consistent manner or have a response linkage relationship during operation. For example, if the conduction states of two interface modules switch simultaneously at most times, they can be considered to have a strong conduction behavior correlation.

[0048] For example, by analyzing the numerical distribution in the state difference map, we focus on component pairs with small difference values ​​and identify them as potential linkage components. In the specific clustering process, we use a hierarchical similarity integration mechanism to gradually group linkage components with difference values ​​below a set threshold into the same cluster unit, forming multiple cluster units with high consistency in conduction behavior. Each cluster unit is considered a conduction characteristic association group, representing a collection of components with mutually influential states or common evolutionary characteristics during operation. On this basis, we further calculate structural parameters such as the state change path, synchronous switching frequency, and common response period between components within each cluster unit to form a complete conduction characteristic association structure. This conduction characteristic association structure has the dual attributes of nodes (components) and edges (state association relationships), and can describe the propagation path and coupling mode of conduction characteristics within the system.

[0049] Step S303 : Combining various conduction correlation factors in the conduction characteristic correlation structure, multi-dimensional modeling is performed on the conduction characteristic changes of each component to obtain a monitoring indicator set.

[0050] Among them, various types of conduction correlation factors represent a set of parameters used to measure the correlation characteristics of conduction behaviors between components in the conduction characteristic correlation structure, and are used to support subsequent conduction behavior modeling.

[0051] For example, various conduction correlation factors are extracted based on the conduction characteristic correlation structure. These factors include multi-dimensional quantitative factors such as the synchronization coefficient between components, response coupling strength, state transition delay, and behavioral consistency score. These factors are used to quantify the intrinsic connection between different components in their conduction behavior. Subsequently, the conduction state sequence of each component is combined with its positional relationship in the conduction characteristic correlation structure, and modeling is performed from both local and global perspectives. Specifically, local modeling focuses on the conduction characteristics of the component itself, such as state change frequency, amplitude jitter range, and abnormal response duration. Global modeling focuses on the associated behavior between a component and other components, such as whether it switches the conduction state synchronously with other components under the influence of a certain event, and whether there is leading or lagging behavior.

[0052] On this basis, by integrating the above information, a multidimensional set of monitoring indicators reflecting the changes in conduction characteristics is constructed. Each of these monitoring indicators not only has quantitative characteristics, but also embeds structural dependency information. For example, a monitoring indicator can represent the "average conduction stability index of component A in the linked components" or the "state switching consistency score between component B and its associated components"; ultimately, the monitoring indicator set, as output, includes system-level parameters that can characterize the overall picture of conduction characteristic changes.

[0053] In this embodiment, first, based on the differential characteristics of the conduction states of each component in terms of the change amplitude and change direction, the components are subjected to differential calculation processing, thereby establishing a structured state difference map that reflects the conduction state change relationship; secondly, based on the component behavior association relationship in the state difference map, linkage clustering processing is performed, thereby constructing a conduction characteristic association structure for accurately describing the conduction behavior association; thirdly, based on the conduction association factors in the conduction characteristic association structure, multi-dimensional modeling of the conduction characteristic change is implemented to generate a set of monitoring indicators. Based on this, a system closed loop can be achieved from state difference identification, association structure construction to indicator modeling, thereby enhancing the structural expression ability and parameter integrity of the conduction behavior modeling.

[0054] In an exemplary embodiment, according to the drift patterns of the respective monitoring indicators in the monitoring indicator set, the monitoring indicator set is subjected to fusion evolution processing to obtain a trend label set characterizing abnormal evolution trends, including steps S401 to S403.

[0055] Step S401 : Analyze the drift pattern of each monitoring indicator according to the change interval and change direction between the continuous characteristics of each monitoring indicator to obtain the drift characteristic analysis result of each monitoring indicator.

[0056] Among them, the change range and change direction of each monitoring indicator between continuous characteristics represent the numerical change amplitude and change direction reflected by the numerical characteristics in adjacent time periods in the time series of the monitoring indicator, which is used to identify the dynamic trend characteristics of the monitoring indicator during operation. For example, if the value of a monitoring indicator gradually increases by 5 units in 5 consecutive time slices, it shows a positive change direction and a stable growth change range.

[0057] The drift characteristic analysis results represent a set of analysis outputs used to describe information such as the change amplitude, direction, and stability of the monitoring indicator. For example, the analysis structure of "monitoring indicator A shows a moderate decline within a certain time period, the change lasts for 6 cycles, and there is no obvious rebound" can be seen.

[0058] For example, first, each monitoring indicator needs to be organized in chronological order to form a stable time series data structure. Furthermore, a sliding window analysis is performed on the indicator change trend within each time period in the time series data. Specifically, within each analysis window, the change interval of the indicator value, that is, the difference between the previous and next values, is calculated. At the same time, the direction of the continuous change is combined to determine whether it is continuously rising, continuously falling, or oscillating. Based on the determination of the direction of change, the rate of change and the level of stability need to be measured. For example, the standard deviation or slope of change is calculated to reflect the intensity and persistence of the indicator drift. Furthermore, to avoid the interference of short-term fluctuations on the analysis results, the data needs to be moderately smoothed to enhance the components reflecting long-term trends. Based on this, after the above processing is completed, each monitoring indicator can form a set of drift characteristic parameters including the change amplitude, change direction, and change persistence, that is, the drift characteristic analysis results of each monitoring indicator are obtained. These parameters are used to describe the behavior pattern of the corresponding monitoring indicator during operation, that is, whether there are stable offsets, short-term rebounds, cyclical fluctuations, etc.

[0059] Step S402 : performing fusion evolution processing on the monitoring indicator set according to the temporal correlation between the drift characteristic analysis results of each monitoring indicator, and obtaining an evolution model set that characterizes the overall indicator change process.

[0060] Among them, the evolutionary model set is a structured model set formed after the fusion evolutionary modeling of the overall monitoring indicator change process, which is used to express the coordinated change path of the system operation status over time; each evolutionary model in the evolutionary model set corresponds to a specific type of indicator coordinated behavior, such as synchronous rise, synchronous decline, synchronous lagging response, alternating fluctuations, and other evolutionary model categories that describe the coordinated behavior of different indicators.

[0061] For example, first, the drift characteristic analysis results of each monitoring indicator are time-aligned to ensure comparability at the same time point; then, using metrics such as covariance, correlation coefficient, or consistency of change direction, the behavioral trends between different monitoring indicators are analyzed to determine which monitoring indicators exhibit the same-direction drift or causal relationship. Based on this, according to the monitoring indicators that exhibit the same-direction drift or causal relationship, the monitoring indicator groups with high temporal consistency are determined to be merged into an evolutionary unit, and their overall change behavior is modeled; this modeling method not only focuses on the trend changes of individual indicators, but also considers their structural position and dynamic coupling characteristics in the overall operation logic of the system, thereby constructing an evolutionary model set that can reflect the coordinated changes of multi-dimensional trends; among them, each evolutionary model in the evolutionary model set corresponds to a specific type of indicator coordinated behavior, such as multiple monitoring indicators simultaneously experiencing an upward offset during the load increase period, or jointly showing a slow downward trend during the aging stage.

[0062] Step S403 : extracting the change process representing the continuous deviation trend in each monitoring indicator according to the evolution model set, and obtaining a trend label set representing the abnormal evolution trend.

[0063] Among them, the continuous offset trend means that in the time series covered by the evolution model set, a certain monitoring indicator continues to change in the same direction in multiple consecutive time periods and reaches a trend behavior of preset change intensity and time length. It is used to identify long-term deviation states that may represent abnormal evolution, such as "monitoring indicator B is in a slow rising state for 12 consecutive cycles, and the cumulative increase exceeds the upper limit of the normal range."

[0064] For example, within a collection of evolutionary models, the coordinated behavior of indicators within each evolutionary model is analyzed to identify any continuous, unidirectional trend changes with magnitudes exceeding normal ranges. This process involves filtering and extracting these trend changes by setting rules such as change thresholds, duration windows, and minimum trend lengths to identify abnormal evolutionary trends. Furthermore, for each identified abnormal evolutionary trend, its relative position within the evolutionary model is further evaluated to determine whether it represents a structural change node, such as the starting point where multiple monitoring indicators simultaneously shift, or a turning point where a monitoring indicator's state changes from stability to instability. Each confirmed abnormal evolutionary trend is then structured and encapsulated as a trend label. The label includes characteristic parameters such as the indicator number, change direction (increase / decrease), duration, peak shift magnitude, and the components involved. These trend labels are output as analysis results and can be used to quickly determine whether the current system exhibits a specific type of abnormal evolutionary behavior.

[0065] In this embodiment, first, the drift characteristics of the monitoring indicators are analyzed according to the change interval and change direction between continuous characteristics, so as to extract quantitative characteristics reflecting the indicator change behavior, namely the drift characteristic analysis results; secondly, according to the temporal correlation between the drift characteristic analysis results, the monitoring indicator set is fused evolution modeled, so as to construct an evolution model set that characterizes the coordinated changes of the overall operating state of the system; thirdly, according to the evolution model set, the change process showing a continuous offset trend in each monitoring indicator is extracted, so as to generate a trend label set for classifying and identifying abnormal evolution trends. Based on this, the ability to characterize the dynamic evolution laws of the system and the ability to pre-identify abnormal evolution behaviors are enhanced.

[0066] In an exemplary embodiment, based on the temporal correlation between the drift characteristic analysis results of each monitoring indicator, the monitoring indicator set is subjected to fusion evolution processing to obtain an evolution model set that characterizes the overall indicator change process, including steps S501 to S502.

[0067] Step S501 , by identifying the correlation between each monitoring indicator in terms of change direction, change amplitude and response delay, time-aligning the drift characteristic analysis results of each monitoring indicator, and obtaining a time series correlation matrix that characterizes the degree of time series correlation of the monitoring indicators.

[0068] Among them, the timing correlation matrix is ​​a two-dimensional structured matrix used to represent the timing coupling relationship between various monitoring indicators. Each element in it is a structured expression of the timing correlation between two monitoring indicators in terms of change direction, change amplitude, and response delay.

[0069] For example, first, the drift characteristic analysis results of each monitoring indicator are organized into a standardized sequence in chronological order to ensure their synchronization in the time dimension. Then, when comparing each set of monitoring indicators, the degree of temporal correlation between the monitoring indicators is established by matching whether their trend directions are consistent, whether the change amplitudes fluctuate within the same range, and whether there is a triggering relationship between the two. For example, when the upward trend of one monitoring indicator lasts for three cycles and another monitoring indicator shows a similar trend, the response delay relationship between the two is recorded as a temporal correlation degree in terms of response delay. Subsequently, all pairs of related monitoring indicators are represented in a unified matrix, namely the temporal correlation matrix. Each element of the temporal correlation matrix contains the temporal correlation degree between the two monitoring indicators in terms of change direction, change amplitude, and response delay. The temporal correlation matrix is ​​used to quantitatively describe the trend linkage between multiple monitoring indicators and can integrate the scattered single indicator trend information into a collaborative feature set with an internal logical structure.

[0070] Step S502 : by aggregating the indicator collaborative behaviors of the associated monitoring indicators, performing fusion evolution processing on the collaborative change paths between multiple monitoring indicators in the time series correlation matrix, and obtaining an evolution model set that characterizes the overall indicator change process.

[0071] Among them, the collaborative change path represents a temporal structure sequence formed by aggregating and modeling the collaborative behaviors of multiple monitoring indicators on the basis of the temporal correlation between them. It is used to describe the linkage change trend of multiple monitoring indicators during operation. For example, under a certain load disturbance, the three monitoring indicators show an upward trend in turn, forming a collaborative change path of "A rises first, then B rises with a lag, and then C drops briefly".

[0072] For example, first, the associated monitoring indicators in the time series correlation matrix are aggregated. That is, monitoring indicators with high correlation strength in terms of change direction and change amplitude and acceptable response delay are combined into the same evolution unit. Each evolution unit represents a set of key indicator combinations that exhibit synergistic behavior in terms of trend, which may include multiple parameter items that exhibit synergistic responses in the context of events such as load changes, temperature rises, or current fluctuations. Furthermore, after the aggregation is completed, the time series within each evolution unit needs to be structurally modeled. By extracting structural elements such as its change rhythm, synchronization trend, and mutation point distribution, its synergistic change path on the time axis is reconstructed. Furthermore, for the interactions between different evolution units, the triggering sequence, mutual dependence, and possible feedback mechanisms between each synergistic change path are clarified. On this basis, the final output evolution model set is a composite trend model covering multiple key indicator combinations, which structurally reflects the multidimensional synergistic relationships during system operation.

[0073] In this embodiment, first, time alignment is performed based on the correlation between the drift characteristic analysis results of each monitoring indicator in terms of change direction, change amplitude and response delay, so as to construct a time correlation matrix that can quantify the degree of time correlation of the monitoring indicators; secondly, by aggregating the indicator collaborative behavior of the related monitoring indicators, the collaborative change paths between multiple monitoring indicators are fused and evolved in the time correlation matrix, so as to construct a set of evolutionary models that can describe the overall collaborative trend of the system. Based on this, the modeling and integration of the dynamic relationship between multiple indicators can be realized, forming a system-level evolutionary model with time sequence, response logic and structural hierarchy.

[0074] In an exemplary embodiment, based on the trend tag set, the monitoring indicator set is analyzed for evolutionary state with dual constraints of abnormal evolution trajectory and trend tag to obtain a prediction result set for predicting switch component failure, including steps S601 to S603.

[0075] Step S601 : Based on the correspondence between the trend tag set and the monitoring indicator set, the monitoring indicator associated with each trend tag is tracked to obtain a set of abnormal evolution tracks under the trend tag constraint.

[0076] Among them, the set of abnormal evolution trajectories under the constraints of trend labels represents the actual change trajectory sequence extracted from the monitoring indicator set and matching these trend labels based on the characteristic parameters such as change direction, change intensity, and duration defined by the trend labels. It is used to identify the indicator evolution process with significant deviation behavior and trend characteristics.

[0077] For example, first, it is clear which time interval in which set of monitoring indicators each trend label corresponds to, so as to identify the corresponding relationship between the trend label set and the monitoring indicator set; on this basis, the time interval covered by each trend label is scanned indicator by indicator, and the change trajectory of each monitoring indicator in the time interval is extracted, and these change trajectories are compared with the characteristic parameters such as change direction, intensity, continuity, etc. recorded in the trend label, and the abnormal evolution trajectories that overlap in space and match in numerical changes are screened out. Furthermore, after these abnormal evolution trajectories are extracted, they are uniformly classified into the abnormal evolution trajectory set, which not only includes the numerical change path of the specific monitoring indicator, but also encapsulates the logical binding relationship between it and the trend label, so that each abnormal evolution trajectory has an explainable label attribution attribute, and can accurately extract the abnormal change process with structural trend characteristics in the original monitoring data.

[0078] Step S602 : Based on the joint features of the abnormal evolution trajectory set and the trend label set, constraint parsing is performed on the current evolution state of each monitoring indicator to obtain a component evolution state set characterized by the dual constraints.

[0079] Among them, the component evolution state set represents the current state set obtained by combining the joint features in the abnormal evolution trajectory set and the trend label set, and performing a dual constraint analysis on the current trend evolution stage of each monitoring indicator. It is used to express the evolution position and behavior trend of each component in the context of trend continuation.

[0080] For example, first, each abnormal evolution trajectory is compared with the trend label according to its historical change direction, trend amplitude, deviation duration and other parameters to determine whether it conforms to the trend type defined by the trend label in terms of evolution mode. Secondly, the current evolution state of each monitoring indicator is intercepted in real time, and its change rate and direction within a short-term time window are calculated to determine whether the evolution state of the monitoring indicator is within the continuation interval of a certain abnormal evolution trajectory. On this basis, when the trend type constraint defined by the trend label and the continuation characteristic constraint of the corresponding abnormal evolution trajectory are met at the same time, the corresponding evolution state can be classified into the component evolution state set; each item in the component evolution state set not only retains the current numerical state, but also marks the source of its evolution trajectory and the trend label, forming a comprehensive description of the current evolution stage of the component.

[0081] Step S603 : performing prediction analysis on the evolution states that meet the abnormal change trend in the component evolution state set to obtain a prediction result set for predicting switch component failure.

[0082] For example, first, it is necessary to confirm whether each evolutionary state is in a continuous abnormal range, that is, to determine whether it continues the high-risk deviation pattern in the trend label, such as continuous rise leading to overload, long-term decline indicating response attenuation, etc.; for evolutionary states that meet the above-mentioned high-risk deviation pattern, their previous trajectory, trend label information and state change rate are jointly input into the preset prediction model. The prediction model performs fault prediction-level extrapolation calculations on the evolution direction of the corresponding evolutionary state in several future time periods based on the development results of similar historical states. In the final prediction result set, each item contains the target component number, predicted evolution type, expected failure time window and risk level identification.

[0083] In this embodiment, first, trajectory positioning processing is performed based on the correspondence between trend labels and monitoring indicators, so as to extract a set of abnormal evolution trajectories that meet the trend characteristics; secondly, dual constraint analysis is performed based on the joint characteristics of abnormal trajectories and trend labels, so as to accurately obtain a set of component evolution states that characterize the components under the influence of dual constraints; thirdly, predictive analysis processing is performed based on the evolution states that meet the abnormal change trend, so as to obtain a set of fault prediction results with a clear structure and trend basis; based on this, trend-driven state identification and risk extrapolation can be realized, so that the prediction results have the structural advantages of clear trajectory sources and traceable trend characteristics, providing data support and evolution basis for accurate early warning of key components.

[0084] In an exemplary embodiment, predictive analysis is performed on the evolution states that meet the abnormal change trend in the component evolution state set to obtain a prediction result set for predicting switch component failure, including steps S701 to S702.

[0085] Step S701 : performing fault feature recognition on the evolution state that meets the abnormal change trend according to the offset direction, duration and state fluctuation amplitude of each evolution state in the component evolution state set, and obtaining a fault feature recognition result.

[0086] Among them, the fault feature identification result represents a structured identification output used to describe whether the evolution state has fault signs, that is, it is used to support subsequent fault prediction analysis; for example, the fault feature identification result includes "the current evolution state of component A is continuously declining, with high fluctuation amplitude and duration exceeding the threshold", and the evolution state is marked as the fault feature category corresponding to "declining instability".

[0087] Among them, the fault feature category means that the fault modes that may cause system functional abnormalities are divided into several types of fault risk groups based on the identified evolutionary state characteristics, which are used to classify and identify the causes of abnormal operation of different components and conduct predictive modeling. For example, the fault feature category may include continuous overheating, transient shock, unstable conduction, communication loss, etc.

[0088] For example, first, the three core parameters corresponding to each evolutionary state are clarified: the offset direction, duration, and state fluctuation amplitude. The offset direction is used to measure whether the current evolutionary state is offsetting upward, downward, or undergoing repeated changes. The duration is used to measure the duration of the trend reflected by the current evolutionary state. The fluctuation amplitude is used to measure the stability and degree of disturbance of the current evolutionary state during the evolution process. Based on this, by jointly analyzing the above three parameters, the dynamic characteristic profile of each evolutionary state can be constructed, and based on this, it can be judged whether it meets the defined abnormal trend standards. For example, when an evolutionary state shows a long-term unidirectional offset accompanied by violent fluctuations, then the evolutionary state shows a high-risk trend in the characteristic dimension and needs to be marked as a fault state to be evaluated.

[0089] Furthermore, during the actual processing process, a mapping mechanism must be established to map the deviation behavior to known historical fault signature categories. By matching the degree of deviation between the current evolution state and historical samples, the fault signature category to which it is close can be determined. Furthermore, after processing, these results can be organized into fault signature identification results, which may include key fields such as the component identifier of the evolution state, the deviation direction mark, the fluctuation level assessment value, the trend duration interval, and the fault signature category, forming the basic data set for predictive pre-analysis.

[0090] Step S702 : In the fault feature identification result, comprehensively analyze the amplitude mutation feature and the change continuity feature of the corresponding evolution state to obtain a prediction result set for predicting the switch component failure.

[0091] Among them, the amplitude mutation feature refers to the phenomenon of a large numerical jump in the monitoring indicator in a short period of time during the evolution of the component state. It is used to identify the critical behavior characteristics of the rapid transition from a normal fluctuation state to an abnormal state.

[0092] The change continuation feature indicates the length of time and trend consistency that the detection indicator continues to change in the same direction during the component evolution state change process. It is used to determine whether a certain deviation behavior has the continuous evolution feature of developing into a fault.

[0093] For example, the system first quantitatively analyzes the amplitude mutation characteristics of each evolutionary state, calculating its rate of change per unit time and its deviation from the steady-state interval to determine whether it meets the threshold criteria for triggering a fault. Next, a temporal evaluation of the change continuity characteristics is performed, tracking the duration of the corresponding evolutionary state since its first appearance, its trend consistency, and whether there are any behavioral anomalies such as interruptions or reversals, to confirm whether the evolutionary state has the coherence and continuity to develop into a fault. Based on this analysis, the amplitude mutation characteristics and change continuity characteristics are multi-dimensionally integrated, and the fault risk level of the evolutionary state is classified according to preset judgment rules, and the corresponding fault prediction results are generated accordingly.

[0094] Furthermore, the prediction result set will include the fault risk level, expected risk time window, key indicators involved, and reference evolution path corresponding to each suspicious evolution state, which together constitute a structured prediction output at the switch component level. This prediction result set is not only used to ultimately determine whether there is an impending failure event, but also supports risk grading, maintenance resource scheduling, and key location monitoring strategy updates in subsequent fault response processes, thereby converting the qualitative identification of evolutionary states into system-level fault prediction data support.

[0095] In this embodiment, first, fault feature identification and processing are performed based on the offset direction, duration and fluctuation amplitude of the evolution state, so as to achieve systematic classification and risk attribution of the evolution state that meets the abnormal change trend, and obtain the fault feature identification result; secondly, comprehensive analysis and processing are performed based on the amplitude mutation characteristics and change continuation characteristics in the fault feature identification result, so as to output a set of prediction results supported by time series continuity and trend strength. Based on this, early identification and classification prediction of potential faults are achieved, and the structural understanding and pre-intervention capabilities of the risk evolution process are improved.

[0096] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0097] Based on the same inventive concept, embodiments of the present application also provide a system for intelligent monitoring and fault prediction of switches for implementing the aforementioned method for intelligent monitoring and fault prediction of switches. The solution provided by this system is similar to the solution described in the aforementioned method. Therefore, the specific limitations of the embodiments of the intelligent monitoring and fault prediction system for one or more switches provided below can be found in the aforementioned limitations of the intelligent monitoring and fault prediction method for switches, and will not be further elaborated here.

[0098] In an exemplary embodiment, Figure 2 As shown, a switch intelligent monitoring and fault prediction system is provided, including: a monitoring module 201, a trend analysis module 202 and a prediction module 203, wherein: Monitoring module 201 is used to analyze the basic electrical signals generated by each component in the switch at the level of conductivity characteristics by combining feature decoding and state mapping to obtain a set of monitoring indicators; The trend analysis module 202 is used to perform fusion evolution processing on the monitoring indicator set according to the drift pattern of each monitoring indicator in the monitoring indicator set to obtain a trend label set that characterizes the abnormal evolution trend; The prediction module 203 is configured to perform an evolution state analysis on the monitoring indicator set based on the trend tag set, combining the dual constraints of abnormal evolution trajectory and trend tag, to obtain a prediction result set for predicting switch component failure.

[0099] In an exemplary embodiment, the monitoring module 201 is also used to: identify key change trends in signal responses, perform feature decoding processing on the basic electrical signals generated by each component, and obtain a feature information set that characterizes the conductive behavior of the component; construct a response association relationship between features and states, and perform state mapping processing on the conductive state of each component in the feature information set to obtain a state representation set that characterizes the conductive state differences; obtain a matching conductive characteristic association structure based on the conductive state differences represented by the state representation set, and perform multi-dimensional modeling on the conductive characteristic changes of each component based on the conductive characteristic association structure to obtain a monitoring indicator set.

[0100] In an exemplary embodiment, the monitoring module 201 is also used to: perform difference calculation processing on each component in the state representation set based on the difference characteristics of the conduction state between each component in the change amplitude and change direction, and obtain a state difference map that characterizes the change relationship of the conduction state of the components; cluster the components that have correlation in conduction behavior based on the state difference map, and obtain a conduction characteristic correlation structure that characterizes the correlation of conduction behavior; combine various conduction correlation factors in the conduction characteristic correlation structure to perform multi-dimensional modeling on the conduction characteristic changes of each component to obtain a set of monitoring indicators.

[0101] In an exemplary embodiment, the trend analysis module 202 is also used to: analyze the drift pattern of each monitoring indicator according to the change interval and change direction between the continuous characteristics of each monitoring indicator to obtain the drift characteristic analysis results of each monitoring indicator; perform fusion evolution processing on the monitoring indicator set according to the temporal correlation relationship between the drift characteristic analysis results of each monitoring indicator to obtain an evolution model set that characterizes the overall indicator change process; based on the evolution model set, extract the change process that characterizes the continuous offset trend in each monitoring indicator to obtain a trend label set that characterizes the abnormal evolution trend.

[0102] In an exemplary embodiment, the trend analysis module 202 is also used to: identify the correlation between each monitoring indicator in terms of change direction, change amplitude and response delay, time-align the drift characteristic analysis results of each monitoring indicator, and obtain a time series correlation matrix that characterizes the degree of time series correlation of the monitoring indicators; by aggregating the indicator collaborative behavior of related monitoring indicators, the collaborative change paths between multiple monitoring indicators are fused and evolved in the time series correlation matrix to obtain a set of evolutionary models that characterize the overall indicator change process.

[0103] In an exemplary embodiment, the prediction module 203 is also used to: locate the trajectory of the monitoring indicators associated with each trend tag according to the correspondence between the trend tag set and the monitoring indicator set, and obtain a set of abnormal evolution trajectories under the trend tag constraints; perform constraint parsing processing on the current evolution state of each monitoring indicator according to the joint characteristics of the abnormal evolution trajectory set and the trend tag set, and obtain a set of component evolution states that characterize the components under the influence of dual constraints; perform predictive analysis on the evolution states that meet the abnormal change trend in the component evolution state set, and obtain a set of prediction results for predicting switch component failures.

[0104] In an exemplary embodiment, the prediction module 203 is also used to: identify fault characteristics of evolution states that meet abnormal change trends based on the offset direction, duration, and state fluctuation amplitude of each evolution state in the component evolution state set, and obtain fault characteristic identification results; in the fault characteristic identification results, comprehensively analyze the amplitude mutation characteristics and change continuation characteristics of the corresponding evolution state to obtain a prediction result set for predicting switch component failures.

[0105] Each module in the intelligent switch monitoring and fault prediction system described above can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor within a computer device in hardware form, or stored in a computer device memory in software form, allowing the processor to call and execute the corresponding operations of each module.

[0106] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in any of the above embodiments when executing the computer program.

[0107] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in any of the above embodiments are implemented.

[0108] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0109] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0110] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for intelligent monitoring and fault prediction of a switch, characterized in that: The method comprises: By combining feature decoding with state mapping, the basic electrical signals generated by each component in the switch are analyzed at the level of conduction characteristics to obtain a set of monitoring indicators. According to the drift pattern of each monitoring indicator in the monitoring indicator set, the monitoring indicator set is subjected to fusion evolution processing to obtain a trend label set representing an abnormal evolution trend; According to the trend tag set, the monitoring indicator set is subjected to an evolution state analysis combining dual constraints of abnormal evolution trajectory and trend tag, to obtain a prediction result set for predicting switch component failure.

2. The method according to claim 1, characterized in that The combined approach of feature decoding and state mapping analyzes the basic electrical signals generated by each component in the switch at the level of conduction characteristics to obtain a set of monitoring indicators, including: By identifying key changing trends in signal responses, the basic electrical signals generated by each component are decoded and processed to obtain a set of characteristic information that characterizes the component's conduction behavior. By constructing a response association relationship between features and states, a state mapping process is performed on the conduction state of each component in the feature information set to obtain a state representation set that characterizes the conduction state differences; A matching conduction characteristic association structure is obtained according to the conduction state differences represented by the state representation set, and a multi-dimensional model is performed on the conduction characteristic changes of each component based on the conduction characteristic association structure to obtain a monitoring indicator set.

3. The method according to claim 2, characterized in that The matching conduction characteristic association structure is obtained based on the conduction state differences represented by the state representation set, and the conduction characteristic changes of each component are multi-dimensionally modeled based on the conduction characteristic association structure to obtain a monitoring indicator set, including: Based on the difference characteristics of the conduction states of the components in terms of change amplitude and change direction, a difference calculation process is performed on each component in the state representation set to obtain a state difference map representing the change relationship of the conduction states of the components; Clustering components that have correlation in conduction behavior according to the state difference map to obtain a conduction characteristic correlation structure that characterizes the correlation in conduction behavior; By combining various conduction correlation factors in the conduction characteristic correlation structure, a multi-dimensional model is performed on the conduction characteristic change of each component to obtain a monitoring index set.

4. The method according to claim 1, wherein The step of performing fusion evolution processing on the monitoring indicator set according to the drift pattern of each monitoring indicator in the monitoring indicator set to obtain a trend label set characterizing an abnormal evolution trend includes: According to the change interval and change direction of each monitoring indicator between continuous characteristics, the drift pattern of each monitoring indicator is analyzed to obtain the drift characteristic analysis results of each monitoring indicator; According to the temporal correlation between the drift characteristic analysis results of each monitoring indicator, the monitoring indicator set is subjected to fusion evolution processing to obtain an evolution model set that characterizes the overall indicator change process; According to the evolution model set, the change process representing the continuous deviation trend in each monitoring indicator is extracted to obtain a trend label set representing the abnormal evolution trend.

5. The method according to claim 4, characterized in that According to the temporal correlation between the drift characteristic analysis results of each monitoring indicator, the monitoring indicator set is subjected to fusion evolution processing to obtain an evolution model set that characterizes the overall indicator change process, including: By identifying the correlation between the change direction, change amplitude and response delay of each monitoring indicator, the drift characteristic analysis results of each monitoring indicator are time-aligned to obtain the time series correlation matrix that represents the degree of time series correlation of the monitoring indicators; By aggregating the indicator collaborative behaviors of related monitoring indicators, the collaborative change paths between multiple monitoring indicators are fused and evolved in the time series association matrix to obtain a set of evolutionary models that characterize the overall indicator change process.

6. The method according to claim 1, characterized in that The step of performing an evolution state analysis on the monitoring indicator set based on the trend tag set in combination with the dual constraints of abnormal evolution trajectory and trend tag to obtain a prediction result set for predicting switch component failure includes: According to the correspondence between the trend tag set and the monitoring indicator set, the monitoring indicator associated with each trend tag is tracked and the abnormal evolution trajectory set under the trend tag constraint is obtained; According to the joint features of the abnormal evolution trajectory set and the trend label set, the current evolution state of each monitoring indicator is subjected to constraint analysis to obtain a component evolution state set characterized by the dual constraints; Predictive analysis is performed on the evolution states that meet the abnormal change trend in the component evolution state set to obtain a prediction result set for predicting switch component failure.

7. The method according to claim 6, characterized in that The predictive analysis of the evolution states that meet the abnormal change trend in the component evolution state set to obtain a prediction result set for predicting switch component failures includes: According to the offset direction, duration, and state fluctuation amplitude of each evolution state in the component evolution state set, fault feature identification is performed on the evolution state that meets the abnormal change trend to obtain a fault feature identification result; In the fault feature identification results, a comprehensive analysis is performed on the amplitude mutation features and change continuity features of the corresponding evolution state to obtain a prediction result set for predicting switch component failures.

8. An intelligent monitoring and fault prediction system for a switch, characterized in that: The system comprises: The monitoring module is used to analyze the basic electrical signals generated by each component in the switch at the level of conductivity characteristics by combining feature decoding and state mapping to obtain a set of monitoring indicators; A trend analysis module, configured to perform fusion evolution processing on the monitoring indicator set according to the drift pattern of each monitoring indicator in the monitoring indicator set, and obtain a trend label set representing an abnormal evolution trend; The prediction module is used to perform an evolution state analysis of the monitoring indicator set based on the trend label set, combining the dual constraints of abnormal evolution trajectory and trend label, to obtain a prediction result set for predicting switch component failure.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images