An intelligent maintenance system and method for a water power station isolation fence based on an internet of things

Abstract

The application discloses a kind of water power station isolation fence intelligent maintenance system and method based on Internet of Things, it is related to equipment intelligent operation and maintenance technical field, this method includes: constructing function failure transmission tree;Function failure transmission tree includes top event, intermediate event and bottom event;Based on historical maintenance record, construct defect feature library, the running state data of the sensor of each fence node is acquired in real time;Running state data is compared with defect feature library, if it is identified that node enters the certain stage of state transition path, then the prediction failure probability and residual safety time that it develops into corresponding bottom event are analyzed;If it is identified that node data is abnormal, then it is matched corresponding bottom event;According to prediction failure probability or corresponding bottom event, based on function failure transmission tree, the comprehensive probability of top event occurrence is dynamically analyzed;According to comprehensive probability and residual safety time, generate early warning information.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance technology, specifically to an intelligent maintenance system and method for isolation fences in hydropower stations based on the Internet of Things. Background Technology

[0002] Hydropower station maintenance isolation fences are crucial facilities for ensuring the safety of maintenance personnel and equipment. Currently, power stations such as Xiluodu generally use traditional post-type fences, whose deployment and retrieval rely entirely on manual handling, resulting in low efficiency. More importantly, these fences only provide physical isolation and lack proactive safety warning capabilities. Personnel climbing over, fence collapse, or equipment malfunctions cannot be detected and alerted in real time, posing significant safety hazards. With the advancement of smart power plant construction, higher demands are being placed on the intelligence of isolation fences.

[0003] Existing technological improvements largely focus on adding simple sensors to fences to achieve local audible and visual alarms, or on remote status transmission via networking. However, these solutions still have fundamental limitations: they can only provide reactive alarms for clearly occurring faults, failing to identify and predict the gradual process of equipment performance degradation in its early stages; furthermore, alarm information is isolated, making it impossible to assess the actual impact of a single device failure on the overall security isolation function of the fence system. This results in maintenance work remaining in a fault-driven, reactive repair mode, lacking foresight and unable to quantify repair priorities. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent maintenance system and method for hydropower station isolation fences based on the Internet of Things, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a smart maintenance method for a hydropower station isolation fence based on the Internet of Things, the method comprising: S100. Construct a functional failure propagation tree; the functional failure propagation tree includes a top event, intermediate events and a bottom event; the top event is the fence security isolation function failure event, the intermediate events are the sub-functional failure events obtained by decomposing the top event into a hierarchical function, and the bottom event is the equipment node failure event that causes the intermediate event to occur. S200. Based on historical maintenance records, a defect feature library is constructed. The defect feature library includes equipment defect features corresponding to bottom events, the initial conditional probability of bottom events causing their direct upper-level events, and the state transition paths of nodes from the normal state to each bottom event. The state transition paths include ordered sub-health features and typical time patterns. S300: Real-time acquisition of sensor status data for each fence node; S400. Compare the running status data with the defect feature library. If a node is identified as entering a certain stage of the state transition path, analyze the predicted failure probability and remaining safe time of its development into the corresponding bottom event. If a node is identified as having abnormal data, match it with the corresponding bottom event. Based on the predicted failure probability or the corresponding bottom event, dynamically analyze the comprehensive probability of the top event occurring based on the functional failure propagation tree. S500 assesses the security of the fence system based on the comprehensive probability and the predefined risk range of the remaining safe time, generates early warning information that includes the immediate risk level and expected risk trend, and pushes the early warning information and the corresponding fault node location to the user platform.

[0006] According to the above scheme, the intermediate event is the top event, and one or more sub-function failure events are obtained by hierarchical functional decomposition. The sub-function failure events include physical obstruction function failure events, warning and alarm function failure events, and status monitoring function failure events. The bottom event is a failure event of one or more device nodes that directly triggers the intermediate event. The device node failure event is associated with the tape, infrared beam detector, audible and visual alarm and communication module that make up the isolation fence.

[0007] According to the above scheme, step S200 includes: S210. Based on historical maintenance records, extract at least one equipment defect feature associated with each bottom event; S220. Through statistical analysis, determine the initial conditional probability of each base event triggering the occurrence of its directly corresponding intermediate event; S230. For each bottom event, analyze at least one state transition path from the normal working state of a node in the functional failure propagation tree to the fault state represented by that bottom event. The state transition path consists of multiple sub-health features arranged in chronological or logical order. The ordered sub-health features are non-faulty but abnormal operating features that gradually appear during the evolution of the equipment node from the normal state to the fault state. Each sub-health feature is arranged according to the order of its appearance or logical progression to form a unique state transition path. It is also associated with a typical time pattern that represents the typical time range from the appearance of the first sub-health feature to the occurrence of the fault. The typical time pattern is a fixed time range obtained based on a large amount of historical evolution data, which accurately reflects the time pattern of the node from the appearance of the first sub-health feature to the final occurrence of the fault.

[0008] According to the above scheme, the operating status data of the device nodes associated with the events at the bottom of the functional failure propagation tree are periodically collected by the wireless sensor network deployed at each fence node. The operational status data includes the working status signal and signal strength of the infrared beam detector, the trigger status and battery voltage of the audible and visual alarm, and the network connection status and signal quality of the communication module.

[0009] According to the above scheme, step S400 includes: S410. Match and compare the real-time collected operating status data with the equipment defect features and sub-health feature sequences in the defect feature library. The matching and comparison adopts the feature parameter matching degree judgment method, and a preset matching degree threshold is set. When the matching degree between the real-time operating status data and a certain feature in the defect feature library reaches or exceeds the threshold, it is judged as a successful match. S420. If the running status data matches the sub-health characteristic sequence of the current stage in a certain state transition path, it is determined that the corresponding node has entered the current stage of the state transition path. Based on the typical time pattern of the state transition path, the predicted failure probability and remaining safe time of the node developing into the corresponding bottom event of the path are analyzed. The predicted failure probability is calculated based on the time difference between the current stage and the failure endpoint, the number and intensity of the sub-health characteristics that have appeared, and the typical time pattern. The remaining safe time is the remaining unevolved time length in the typical time pattern, that is, the expected time from the current stage to the occurrence of the failure. S430. If the operating status data matches any equipment defect feature in the defect feature library, the corresponding node data is determined to be abnormal, and the node is matched to the underlying event corresponding to the equipment defect feature. S440. Based on the predicted failure probability or the corresponding bottom event, and according to the logical association and probability transmission relationship between the bottom event, intermediate event and top event in the functional failure transmission tree, dynamically analyze the comprehensive probability of the top event occurring. The logical association and probability transmission relationship is as follows: the probability of the bottom event occurring is transmitted upward to the corresponding intermediate event. The probability of the intermediate event occurring corresponding to multiple bottom events is logically aggregated according to the function of the top event to form the comprehensive probability of the top event occurring. The aggregation method is based on the importance weight of each intermediate event to the top event.

[0010] According to the above scheme, step S500 includes: S510. The comprehensive probability and remaining safe time obtained from dynamic analysis are compared with multiple predefined risk assessment intervals. The multiple predefined risk assessment intervals are divided according to the safety level requirements of the hydropower station fence, the degree of historical failure loss and operation and maintenance response capability. Each interval corresponds to a clear risk level standard, and the interval division has a fixed threshold and no fuzzy boundaries. S520. Based on the risk assessment intervals of the comprehensive probability and the remaining safe time, analyze the immediate risk level and expected risk trend level corresponding to the current safety status of the fence system. The immediate risk level represents the degree of risk of the current safety status of the fence system, and the expected risk trend level represents the direction of future risk changes predicted based on the remaining safe time. S530 generates early warning information based on the real-time risk level and the expected risk trend level. The early warning information includes the risk level, fault node identifier, fault type, predicted fault time, remaining safe time, and preliminary handling suggestions. The early warning information and the location information of the underlying event or node corresponding to the early warning information are pushed to the user platform.

[0011] An intelligent maintenance system for the isolation fence of a hydropower station based on the Internet of Things (IoT) includes: a knowledge base module, a data acquisition module, an intelligent diagnosis module, and an early warning generation module. The knowledge base module is used to construct a functional failure propagation tree and a defect feature library. The functional failure propagation tree includes the top event representing the failure of the fence security isolation function, the intermediate events of the sub-functional failures obtained by decomposing the top event step by step, and the bottom events of the equipment node failures that cause the intermediate events. The defect feature library includes the equipment defect features corresponding to each bottom event, the initial conditional probability of each bottom event causing its direct upper-level event, and the state transition paths of nodes from the normal state to each bottom event, which contain ordered sub-health features and typical time patterns. The data acquisition module is used to periodically collect the operating status data of the device nodes that constitute the base event; the operating status data includes the working status parameters of the infrared beam detector, the audible and visual alarm, and the communication module; The intelligent diagnostic module is used to match and compare the operating status data with the defect feature library; if a node is identified to have entered a certain stage of the state transition path, it analyzes the predicted failure probability and remaining safe time of the corresponding bottom event; if a node data is identified as abnormal, it matches the corresponding bottom event; and based on the predicted failure probability or the corresponding bottom event, it dynamically analyzes the comprehensive probability of the top event occurring based on the functional failure propagation tree. The early warning generation module is used to assess the security of the fence system based on the comprehensive probability and the predefined risk range in which the remaining safe time is located, and generate early warning information that includes the immediate risk level and expected risk trend.

[0012] According to the above scheme, the knowledge base module includes a failure tree unit, a feature extraction unit, and a migration path unit. The failure tree unit is used to construct a functional failure propagation tree and define the top event, intermediate events, and bottom events. The feature extraction unit is used to extract equipment defect features corresponding to each bottom event based on historical maintenance records, and to determine the initial conditional probability of each bottom event triggering its direct upper-level event through statistical analysis. The migration path unit is used to analyze the process of a node evolving from a normal working state to the fault state represented by the bottom event for each bottom event, and to summarize the state migration path composed of ordered sub-health features and associated with typical time patterns.

[0013] According to the above scheme, the intelligent diagnostic module includes a data matching unit, a status determination unit, and a probability analysis unit. The data matching unit is used to match and compare the real-time collected operating status data with the equipment defect features and sub-health feature sequences in the defect feature library. The status determination unit is used to make a determination based on the matching results. If it matches the sub-health feature sequence of a specific stage in a certain state transition path, it determines that the node has entered that stage of the path and analyzes and predicts the failure probability and remaining safe time based on the typical time pattern of the path. If it matches the equipment defect features, it determines that the node data is abnormal and matches it to the corresponding bottom event. The probability analysis unit is used to dynamically analyze the comprehensive probability of the top event based on the predicted failure probability or the matched bottom event, according to the logical association and probability transmission relationship of events in the functional failure transmission tree.

[0014] According to the above scheme, the early warning generation module includes an interval comparison unit, a risk determination unit, and an early warning synthesis unit. The interval comparison unit is used to compare the comprehensive probability and remaining safe time with multiple predefined risk assessment intervals. The risk determination unit is used to determine the immediate risk level and the expected risk trend level based on the risk assessment interval in which the comprehensive probability and remaining safe time are located. The early warning synthesis unit is used to generate early warning information based on the immediate risk level and the expected risk trend level, and associate it with the corresponding fault node location information for push notification.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention constructs an equipment degradation model consisting of ordered sub-health characteristics and typical time patterns; identifies the early stage of equipment performance degradation, predicts the probability of it developing into functional failure and the remaining safe time, and issues an early warning before the failure occurs, thereby improving the initiative and timeliness of safety protection. 2. This invention integrates the functional failure propagation tree with real-time diagnostic data to dynamically calculate the comprehensive probability that an anomaly in a single device node will cause the entire fence system's safety functions to fail, thereby achieving quantitative and dynamic assessment of the overall safety status of the fence system. 3. The early warning information output by this invention includes both the immediate risk level and the expected risk trend level, informing maintenance personnel of the urgency of the current risk and its future trend, thereby improving decision-making efficiency. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the structure of an intelligent maintenance system and method for a hydropower station isolation fence based on the Internet of Things according to the present invention; Figure 2 This is a schematic diagram of the structure of an intelligent maintenance system and method for a hydropower station isolation fence based on the Internet of Things according to the present invention; Figure 3This is a schematic diagram of an isolation fence for a smart maintenance system and method for a hydropower station based on the Internet of Things, according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example: Figures 1-3 As shown, the present invention provides a technical solution: an intelligent maintenance method for a hydropower station isolation fence based on the Internet of Things, the method comprising the following steps: S100. Construct a functional failure propagation tree; the functional failure propagation tree includes a top event, intermediate events and a bottom event; the top event is the fence security isolation function failure event, the intermediate events are the sub-functional failure events obtained by decomposing the top event into a hierarchical function, and the bottom event is the equipment node failure event that causes the intermediate event to occur. Specifically, intermediate events are one or more sub-function failure events obtained by decomposing the top event into a hierarchical function. These sub-function failure events include physical barrier function failure events, warning alarm function failure events, and status monitoring function failure events. Bottom events are one or more equipment node failure events that directly trigger intermediate events. These equipment node failure events are associated with the tape, infrared beam detectors, audible and visual alarms, and communication modules that constitute the isolation fence. For example: Top event T: The fence security isolation function is completely disabled; Based on the fence's functional structure, the top event T is decomposed into three independent intermediate events: M1: Physical obstruction function failure, such as a broken or unstretched conveyor belt; M2: Warning alarm function failure, such as a malfunctioning sound and light alarm or a faulty infrared beam, resulting in no alarm being triggered when someone enters; M3: Status monitoring function failure, such as a malfunctioning 4G communication module, resulting in the inability to upload status information. Intermediate events M1, M2 and M3 are connected to the top event T through an OR logic gate, meaning that if any intermediate event occurs, the top event T will occur. For each intermediate event, identify the underlying event that caused it: The underlying event B11 that leads to intermediate event M1: physical breakage of the fabric tape; The underlying event B21 that leads to intermediate event M2: contamination of the optical lens of the infrared beam detector causing signal attenuation; the underlying event B22: depletion of the battery in the audible and visual alarm. The underlying event B31 that leads to intermediate event M3: 4G communication module SIM card failure; The bottom event and the intermediate event are connected by an OR gate; This forms a functional failure propagation tree containing 1 top event, 3 intermediate events, and 4 bottom events; this is just an example and is not a limitation.

[0019] S200. Based on historical maintenance records, a defect feature library is constructed. The defect feature library includes equipment defect features corresponding to bottom events, the initial conditional probability of bottom events causing their direct upper-level events, and the state transition paths of nodes from the normal state to each bottom event. The state transition paths include ordered sub-health features and typical time patterns. Specifically, step S200 includes: S210. Based on historical maintenance records, extract at least one equipment defect feature associated with each bottom event; For example: Extract the equipment status features directly corresponding to each basic event from historical maintenance work orders; for basic event B21 infrared beam signal attenuation, its equipment defect feature F_B21 is defined as: signal strength value < threshold S_th; for basic event B22 audible and visual alarm battery depletion, its equipment defect feature F_B22 is defined as: battery voltage < threshold V_th; for basic event B31 communication module failure, its equipment defect feature F_B31 is defined as: network signal quality RSRP < threshold R_th and duration > T_th; thresholds S_th, V_th, R_th and T_th are pre-set threshold values ​​based on equipment manufacturer technical specifications, historical normal operation data statistical characteristics or industry safety standards; S220. Through statistical analysis, determine the initial conditional probability of each base event triggering the occurrence of its directly corresponding intermediate event; For example: In the past year, there were a total of 10 B21 events, of which 8 directly caused the M2 alarm function to fail; then the initial conditional probability of the base event B21 triggering the intermediate event M2 is P(M2|B21) = 8 / 10 = 0.8; successively, we get P(M2|B22) = 0.9, P(M3|B31) = 0.95, P(M1|B11) = 0.99; this is just an example and is not a limitation. S230. For each bottom event, analyze at least one state transition path from the normal working state of a node in the functional failure propagation tree to the fault state represented by that bottom event. The state transition path consists of multiple sub-health features arranged in chronological or logical order. The ordered sub-health features are non-faulty but abnormal operating features that gradually appear during the evolution of the equipment node from the normal state to the fault state. Each sub-health feature is arranged according to the order of its appearance or logical progression to form a unique state transition path. It is also associated with a typical time pattern that represents the typical time range from the appearance of the first sub-health feature to the occurrence of the fault. The typical time pattern is a fixed time range obtained based on a large amount of historical evolution data, which accurately reflects the time pattern of the node from the appearance of the first sub-health feature to the final occurrence of the fault. For example, taking the battery depletion of the B22 audible and visual alarm as an example, the sub-health characteristic sequence is as follows: By analyzing a large amount of historical battery voltage data, it was found that the degradation process exhibits regularity; three ordered sub-health characteristics are defined: H1: The battery voltage first and continuously drops below 11.5V from the nominal value of 12V; H2: The battery voltage continuously drops below 11.0V, and the daily rate of decline accelerates; H3: The battery voltage fluctuates drastically around 10.5V; Typical time pattern: Statistical analysis shows that from the occurrence of H1 to the eventual occurrence of the equipment defect characteristic F_B22 failure, the time range is usually concentrated between 20 and... The typical time pattern T_B22 is set as the interval [20 days, 30 days] within 30 days. The typical time pattern is obtained by statistical analysis of a large number of time samples of similar fault evolution processes in history, such as calculating the mean and determining a time interval with high confidence. The statistical modeling of the evolution process is carried out using methods such as survival analysis. A state transition path is formed for the bottom event B22, where the battery of the audible and visual alarm is exhausted: normal state - H1 - H2 - H3 - fault state (B22), and associated with the time pattern T_B22. This is only an example and is not a limitation.

[0020] S300: Real-time acquisition of sensor status data for each fence node; Specifically, through a wireless sensor network deployed at each fence node, the operating status data of the device nodes associated with the events at the bottom of the functional failure propagation tree are periodically collected; the operating status data includes the working status signal and signal strength of the infrared beam detector, the trigger status and battery voltage of the audible and visual alarm, and the network connection status and signal quality of the communication module. For example: The wireless sensor network collects data every 5 minutes; at a certain moment, the following data is collected: the infrared beam signal strength of node 05 is 25% (threshold is 30%); the battery voltage of the audible and visual alarm of node 05 is 11.0V; the RSRP of the communication module of node 05 is -85dBm; this is only an example and is not a limitation; the collection cycle is dynamically configured according to network load, power supply conditions and monitoring requirements.

[0021] S400. Compare the running status data with the defect feature library. If a node is identified as entering a certain stage of the state transition path, analyze the predicted failure probability and remaining safe time of its development into the corresponding bottom event. If a node is identified as having abnormal data, match it with the corresponding bottom event. Based on the predicted failure probability or the corresponding bottom event, dynamically analyze the comprehensive probability of the top event occurring based on the functional failure propagation tree. Specifically, step S400 includes: S410. Match and compare the real-time collected operating status data with the equipment defect features and sub-health feature sequences in the defect feature library. The matching and comparison adopts the feature parameter matching degree judgment method, and a preset matching degree threshold is set. When the matching degree between the real-time operating status data and a certain feature in the defect feature library reaches or exceeds the threshold, it is judged as a successful match. For example, the real-time data of node 05, including infrared beam signal strength of 25%, audible and visual alarm battery voltage of 11.0V, and communication module RSRP of -85dBm, is compared with the defect feature database. The comparison algorithm uses Euclidean distance or cosine similarity to calculate the degree of match between the real-time data vector and the feature vector. For example, the degree of match between the battery voltage data and the sub-health feature H2 is calculated. A threshold method is used, with a preset match threshold of 90%. When the calculated similarity is ≥90%, the match is considered successful. In this embodiment, 11.0V is the same as H2 (11.0V), so the match is successful. The comparison algorithm is preset and optimized according to the type of different features and noise level. The match threshold is determined by analyzing the statistical distribution of correct and incorrect matching cases in historical data. S420. If the operational status data matches the sub-health characteristic sequence of the current stage in a certain state transition path, then it is determined that the corresponding node has entered the current stage of the state transition path. Based on the typical time pattern of the state transition path, the predicted failure probability and remaining safe time of the node developing into the corresponding bottom event of the path are analyzed. The predicted failure probability is calculated based on the time difference between the current stage and the failure endpoint, the number and intensity of the sub-health characteristics that have appeared, and the typical time pattern. Specifically, the survival function obtained by fitting historical failure time data is used to calculate the conditional failure probability under a given survival time. The remaining safe time is the remaining unevolved time length in the typical time pattern, that is, the expected time from the current stage to the occurrence of the failure. For example: The battery voltage data of the audible and visual alarm of node 05 successfully matched the sub-health characteristic H2 in the state transition path of event B22; confirming that node 05 has entered the H2 stage of path B22; Based on the evolved situation, i.e., H1 and H2 have already occurred, and the remaining time, calculate the probability of a failure occurring within a future time period; for example, using conditional probability calculation based on historical survival analysis; the formula is: P(failure | survived to stage H2, future time Δt) = 1 - S(t_H2 + Δt) / S(t_H2); where S(t) is the survival function from the starting point to time t, obtained by fitting from historical data, and t_H2 is the time to enter stage H2; if the calculated failure probability within the next 7 days is 30%, then the predicted failure probability P_f = 0.3; the survival function S(t) is obtained by fitting historical failure time data using parametric models or non-parametric methods; this is only an example and is not a limitation; S430. If the operating status data matches any equipment defect feature in the defect feature library, the corresponding node data is determined to be abnormal, and the node is matched to the underlying event corresponding to the equipment defect feature. For example: if the infrared beam signal strength of node 05 is found to be 25%, which is lower than the threshold of 30%, it is directly matched to the equipment defect feature F_B21; if node 05 is also found to have data anomalies, it is directly matched to the event B21; direct matching means that the real-time data meets the sufficient conditions for fault determination. This is only an example and is not a limitation. S440. Based on the predicted failure probability or the corresponding bottom event, and according to the logical association and probability propagation relationship between the bottom event, intermediate event and top event in the functional failure propagation tree, dynamically analyze the comprehensive probability of the top event occurring. The logical association and probability propagation relationship is as follows: the probability of the bottom event occurs is propagated upward to the corresponding intermediate event. The probability of the intermediate event occurring corresponding to multiple bottom events is logically aggregated according to the function of the top event to form the comprehensive probability of the top event occurring. The aggregation method is based on the importance weight of each intermediate event to the top event. The importance weight is calculated by the analytic hierarchy process or the entropy weight method based on the historical frequency of each intermediate event failure, the average downtime caused, or the severity of the safety consequences. For example: For the basic event B22, input its predicted failure probability P_f(B22)=0.3, which is the probability of it occurring in the near future; For the basic event B21, since the defect feature has been clearly matched, it is considered that B21 has occurred, and its occurrence probability P(B21)=1. According to the functional failure propagation tree logic, both B21 and B22 lead to intermediate event M2 through the OR gate; the probability P(M2) of intermediate event M2 is calculated as follows: P(M2) = 1 - [(1 - P(B21) × P(M2|B21)) × (1 - P_f(B22) × P(M2|B22))]; Substituting the data: P(B21) = 1, P(M2|B21) = 0.8, P_f(B22) = 0.3, P(M2|B22) = 0.9; the probability P(M2) of intermediate event M2 is obtained as P(M2) = 1 - [(1 - 1 × 0.8) × (1 - 0.3 × 0.9)] = 0.854; The top event T is triggered by M1, M2, and M3 via an OR gate. If M1 and M3 are not triggered at this time, i.e., the probability is 0, then the overall probability P(T) of the top event T is P(M2) = 0.854. If weights are considered, such as the importance weight of M2 to the top event being w2 = 0.4, and the sum of the weights of other events being 0.6, and none of them occurring, then P(T) = w2 × P(M2) = 0.4 × 0.854 = 0.3416. The importance weight w_i is obtained through expert scoring, severity analysis based on historical failure consequences, or criticality calculation of each intermediate event in the logic structure of the top event. This embodiment uses direct OR gate logic, i.e., P(T) = 0.854. This is only an example and is not a limitation.

[0022] S500 assesses the security of the fence system based on the comprehensive probability and the predefined risk range of the remaining safe time, generates early warning information that includes the immediate risk level and expected risk trend, and pushes the early warning information and the corresponding fault node location to the user platform. Specifically, step S500 includes: S510. The comprehensive probability and remaining safe time obtained from dynamic analysis are compared with multiple predefined risk assessment intervals. The predefined multiple risk assessment intervals are divided according to the safety level requirements of the hydropower station fence, the degree of historical failure loss and operation and maintenance response capabilities. Each interval corresponds to a clear risk level standard, and the interval division has a fixed threshold and no fuzzy boundaries. When dividing, industry safety standards, past accident analysis reports, and the routine inspection and emergency response cycle of the operation and maintenance team are referenced for quantitative setting. For example: the system predefines risk assessment intervals; such as: comprehensive probability interval: [0,0.3) is low risk, [0.3,0.7) is medium risk, [0.7,1.0] is high risk; remaining safe time interval: (30 days,∞) is long-term, [7 days,30 days] is medium-term, [0,7 days) is short-term; In this embodiment, a comprehensive probability of 0.854 falls within the high-risk range; a remaining safe time of 15 days falls within the medium-term range; the range thresholds and level classifications need to be predefined before system deployment, in conjunction with power plant safety regulations and operation and maintenance experience, and calibrated based on operational feedback; this is only an example and is not intended to be restrictive. S520. Based on the risk assessment intervals of the comprehensive probability and the remaining safe time, analyze the immediate risk level and expected risk trend level corresponding to the current safety status of the fence system. The immediate risk level represents the degree of risk of the current safety status of the fence system, and the expected risk trend level represents the direction of future risk changes predicted based on the remaining safe time. For example: Immediate risk level: Based on the high risk of comprehensive probability, it is judged as a red alert, that is, high risk; Expected risk trend level: Based on the remaining safe time being medium and the predicted failure probability being relatively high at 0.3, the trend is judged as a rapid increase in risk. S530. Based on the real-time risk level and the expected risk trend level, generate early warning information. The early warning information includes the risk level, fault node identifier, fault type, predicted fault time, remaining safe time and preliminary handling suggestions. Push the early warning information and the location information of the underlying event or node corresponding to the early warning information to the user platform. For example: Generate early warning information: {Early Warning ID: ALERT-2025-001, Time: [Timestamp], Node: 05, Immediate Risk: Red Alert, Expected Trend: Risk is rising rapidly, Related Events: B21 (Infrared Attenuation - Occurred), B22 (Battery Failure - Forecasted), Predicted Failure Time Window: [Current Time + 13 Days, Current Time + 17 Days], Remaining Safe Time: Approximately 15 Days, Recommendation: Immediately check the infrared transmitter at node 05 and plan to replace the battery}; This is just an example and is not a limitation.

[0023] This invention provides a technical solution: an intelligent maintenance system for a hydropower station isolation fence based on the Internet of Things. The system includes: a knowledge base module, a data acquisition module, an intelligent diagnosis module, and an early warning generation module. The knowledge base module is used to construct a functional failure propagation tree and a defect feature library. The functional failure propagation tree includes the top event representing the failure of the fence security isolation function, the intermediate events of the sub-functional failures obtained by decomposing the top event step by step, and the bottom events of the equipment node failures that cause the intermediate events. The defect feature library includes the equipment defect features corresponding to each bottom event, the initial conditional probability of each bottom event causing its direct upper-level event, and the state transition paths of nodes from the normal state to each bottom event, which contain ordered sub-health features and typical time patterns. The data acquisition module is used to periodically collect the operating status data of the device nodes that constitute the base event; the operating status data includes the working status parameters of the infrared beam detector, the audible and visual alarm, and the communication module; The intelligent diagnostic module is used to match and compare the operating status data with the defect feature library; if a node is identified to have entered a certain stage of the state transition path, it analyzes the predicted failure probability and remaining safe time of the corresponding bottom event; if a node data is identified as abnormal, it matches the corresponding bottom event; and based on the predicted failure probability or the corresponding bottom event, it dynamically analyzes the comprehensive probability of the top event occurring based on the functional failure propagation tree. The early warning generation module is used to assess the security of the fence system based on the comprehensive probability and the predefined risk range in which the remaining safe time is located, and generate early warning information that includes the immediate risk level and expected risk trend.

[0024] Specifically, the knowledge base module includes a failure tree unit, a feature extraction unit, and a migration path unit. The failure tree unit is used to construct a functional failure propagation tree and define the top event, intermediate events, and bottom events. The feature extraction unit is used to extract equipment defect features corresponding to each bottom event based on historical maintenance records and to determine the initial conditional probability of each bottom event triggering its direct upper-level event through statistical analysis. The migration path unit is used to analyze the process of a node evolving from a normal working state to the fault state represented by the bottom event for each bottom event, and to summarize the state migration path composed of ordered sub-health features and associated with typical time patterns.

[0025] Specifically, the intelligent diagnostic module includes a data matching unit, a status determination unit, and a probability analysis unit. The data matching unit is used to match and compare the real-time collected operating status data with the equipment defect features and sub-health feature sequences in the defect feature library. The status determination unit is used to make a determination based on the matching results. If it matches the sub-health feature sequence of a specific stage in a certain state transition path, it determines that the node has entered that stage of the path and analyzes and predicts the failure probability and remaining safe time based on the typical time pattern of the path. If it matches the equipment defect features, it determines that the node data is abnormal and matches it to the corresponding bottom event. The probability analysis unit is used to dynamically analyze the comprehensive probability of the top event based on the predicted failure probability or the matched bottom event, according to the logical association and probability transmission relationship of events in the functional failure transmission tree.

[0026] Specifically, the early warning generation module includes an interval comparison unit, a risk determination unit, and an early warning synthesis unit. The interval comparison unit compares the overall probability and remaining safe time with multiple predefined risk assessment intervals. The risk determination unit determines the immediate risk level and the expected risk trend level based on the risk assessment interval in which the overall probability and remaining safe time are located. The early warning synthesis unit generates early warning information based on the immediate risk level and the expected risk trend level, and associates it with the corresponding fault node location information for push notification.

[0027] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A smart maintenance method for isolation fences of hydropower stations based on the Internet of Things, characterized in that: The method includes: S100. Construct a functional failure propagation tree; the functional failure propagation tree includes a top event, intermediate events and a bottom event; the top event is a fence security isolation function failure event, the intermediate events are sub-functional failure events obtained by hierarchically decomposing the top event, and the bottom event is a device node failure event that causes the intermediate event to occur. S200. Based on historical maintenance records, construct a defect feature library. The defect feature library includes equipment defect features corresponding to bottom events, the initial conditional probability of bottom events triggering their direct upper-level events, and the state transition paths of nodes from the normal state to each bottom event. The state transition paths include ordered sub-health features and are associated with typical time patterns. S300: Real-time acquisition of sensor status data for each fence node; S400. Compare the operating status data with the defect feature library. If a node is identified as entering a certain stage of the state transition path, analyze the predicted failure probability and remaining safe time of its development into the corresponding bottom event. If a node is identified as having abnormal data, match it with the corresponding bottom event. Based on the predicted failure probability or the corresponding bottom event, dynamically analyze the comprehensive probability of the top event occurring based on the functional failure propagation tree. S500. Based on the comprehensive probability and the predefined risk range in which the remaining safe time is located, assess the security of the fence system, generate early warning information containing the immediate risk level and expected risk trend, and push the early warning information and the corresponding fault node location to the user platform.

2. The intelligent maintenance method for a hydropower station isolation fence based on the Internet of Things as described in claim 1, characterized in that: The intermediate event is one or more sub-function failure events obtained by hierarchically decomposing the top event. The sub-function failure events include physical obstruction function failure events, warning and alarm function failure events, and status monitoring function failure events. The bottom event is one or more device node failure events that directly trigger the intermediate event, and the device node failure events are associated with the tape, infrared beam detector, audible and visual alarm and communication module that constitute the isolation fence.

3. The intelligent maintenance method for the isolation fence of a hydropower station based on the Internet of Things as described in claim 1, characterized in that: Step S200 includes: S210. Based on the historical maintenance records, extract at least one equipment defect feature associated with each of the bottom events; S220. Through statistical analysis, determine the initial conditional probability that each of the base events triggers the occurrence of the intermediate event directly corresponding to it; S230. For each of the bottom events, analyze at least one state transition path from the normal working state of a node in the functional failure propagation tree to the fault state represented by the bottom event; the state transition path consists of multiple sub-health features arranged in chronological or logical order, and is associated with a typical time pattern that represents a typical time range from the appearance of the first sub-health feature to the occurrence of the fault.

4. The intelligent maintenance method for the isolation fence of a hydropower station based on the Internet of Things as described in claim 1, characterized in that: The operating status data of the device nodes associated with the base events in the functional failure propagation tree are periodically collected through a wireless sensor network deployed at each fence node. The operational status data includes the operating status signal and signal strength of the infrared beam detector, the trigger status and battery voltage of the audible and visual alarm, and the network connection status and signal quality of the communication module.

5. The intelligent maintenance method for a hydropower station isolation fence based on the Internet of Things according to claim 1, characterized in that: Step S400 includes: S410. Match and compare the real-time collected operating status data with the equipment defect features and sub-health feature sequences in the defect feature library; S420. If the running status data matches the sub-health characteristic sequence of the current stage in a certain state transition path, it is determined that the corresponding node has entered the current stage of the state transition path, and based on the typical time pattern of the state transition path, the predicted failure probability and remaining safe time of the node developing into the bottom event corresponding to the path are analyzed. S430. If the operating status data matches any device defect feature in the defect feature library, then the corresponding node data is determined to be abnormal, and the node is matched to the underlying event corresponding to the device defect feature. S440. Based on the predicted failure probability or the corresponding bottom event, and according to the logical association and probability transmission relationship between the bottom event, intermediate event and top event in the functional failure propagation tree, dynamically analyze the comprehensive probability of the top event occurring.

6. The intelligent maintenance method for a hydropower station isolation fence based on the Internet of Things according to claim 1, characterized in that: Step S500 includes: S510. Compare the comprehensive probability obtained from dynamic analysis and the remaining safe time with multiple predefined risk assessment intervals; S520. Based on the risk assessment intervals of the comprehensive probability and the remaining safety time, analyze the immediate risk level and expected risk trend level corresponding to the current safety status of the fence system. S530. Based on the real-time risk level and the expected risk trend level, generate early warning information and push the early warning information and the location information of the corresponding event or node to the user platform.

7. An intelligent maintenance system for a hydropower station isolation fence based on the Internet of Things, characterized in that: The system includes: a knowledge base module, a data acquisition module, an intelligent diagnosis module, and an early warning generation module; The knowledge base module is used to construct a functional failure propagation tree and a defect feature library. The functional failure propagation tree includes a top event characterizing the failure of the fence security isolation function, intermediate events of sub-functional failures obtained by hierarchically decomposing the top event, and bottom events of equipment node failures that cause the intermediate events. The defect feature library includes equipment defect features corresponding to each bottom event, the initial conditional probability of each bottom event triggering its direct upper-level event, and state transition paths of nodes migrating from the normal state to each bottom event, containing ordered sub-health features and typical time patterns. The data acquisition module is used to periodically collect the operating status data of the device nodes that constitute the bottom event; the operating status data includes the working status parameters of the infrared beam detector, the audible and visual alarm, and the communication module. The intelligent diagnostic module is used to match and compare the operating status data with the defect feature library; if a node is identified to have entered a certain stage of the state transition path, the module analyzes the predicted failure probability and remaining safe time of the corresponding bottom event; if a node is identified as having abnormal data, the module matches it with the corresponding bottom event; and based on the predicted failure probability or the corresponding bottom event, the module dynamically analyzes the comprehensive probability of the top event occurring based on the functional failure propagation tree. The early warning generation module is used to assess the security of the fence system based on the comprehensive probability and the predefined risk range in which the remaining safe time is located, and generate early warning information that includes the immediate risk level and expected risk trend.

8. The intelligent maintenance system for a hydropower station isolation fence based on the Internet of Things as described in claim 7, characterized in that: The knowledge base module includes a failure tree unit, a feature extraction unit, and a migration path unit; The failure tree unit is used to construct the functional failure propagation tree and define the top event, intermediate events and bottom event; The feature extraction unit is used to extract equipment defect features corresponding to each bottom event based on historical maintenance records, and to determine the initial conditional probability of each bottom event triggering its direct upper-level event through statistical analysis. The migration path unit is used to analyze the process of a node evolving from a normal working state to the fault state represented by the bottom event for each bottom event, and to summarize the state migration path composed of ordered sub-health characteristics and associated with typical time patterns.

9. The intelligent maintenance system for a hydropower station isolation fence based on the Internet of Things as described in claim 7, characterized in that: The intelligent diagnostic module includes a data matching unit, a status determination unit, and a probability analysis unit. The data matching unit is used to match and compare the real-time collected operating status data with the equipment defect features and sub-health feature sequences in the defect feature library. The state determination unit is used to make a determination based on the matching result. If it matches the sub-healthy feature sequence of a specific stage in a certain state transition path, the node is determined to enter that stage of the path, and the failure probability and remaining safe time are analyzed and predicted based on the typical time pattern of the path. If it matches the characteristics of a device defect, the node data is determined to be abnormal and matched to the corresponding bottom event; The probability analysis unit is used to dynamically analyze the comprehensive probability of the occurrence of the top event based on the predicted failure probability or the matched corresponding bottom event, according to the logical association and probability propagation relationship of events in the functional failure propagation tree.

10. The intelligent maintenance system for a hydropower station isolation fence based on the Internet of Things as described in claim 7, characterized in that: The early warning generation module includes an interval comparison unit, a risk determination unit, and an early warning synthesis unit; The interval comparison unit is used to compare the comprehensive probability and the remaining safe time with multiple predefined risk assessment intervals; The risk determination unit is used to determine the immediate risk level and the expected risk trend level based on the risk assessment interval in which the comprehensive probability and the remaining safe time are located. The early warning synthesis unit is used to generate early warning information based on the real-time risk level and the expected risk trend level, and associate it with the corresponding fault node location information for push notification.

Images