Fault early warning and intelligent regulation and control system for electromechanical equipment of expressway

Abstract

The invention relates to the technical field of highway management, in particular to a highway electromechanical equipment fault early warning and intelligent regulation and control system which comprises a fault feature extraction and trend modeling module, a multi-source information fusion and fault diagnosis module, a fault processing strategy generation module and a closed-loop feedback and emergency processing module. The system constructs a fault development trend model through data acquisition, feature extraction and clustering analysis, carries out fault diagnosis by fusing multi-source information, generates a fault processing instruction, realizes closed-loop feedback and emergency processing, realizes the conversion from passive maintenance to active prevention, and realizes fault diagnosis through fault trend prediction and evolution rule modeling. Potential risks can be identified in the fault germination stage, preventive measures are taken in advance, and sudden faults of equipment are remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of highway management, and particularly to a fault early warning and intelligent regulation system for highway electromechanical equipment. Background Art

[0002] Highway electromechanical equipment is an important infrastructure to ensure the safe and efficient operation of highways, including various types such as monitoring equipment, communication equipment, toll collection equipment, lighting equipment, etc. These devices are exposed to complex external environments for a long time, and have heavy workloads, with a high probability of failure.

[0003] The existing highway electromechanical equipment management systems mainly adopt a passive management mode of periodic inspection and post-fault maintenance. This mode has many drawbacks: on the one hand, maintenance is carried out only after equipment failure, which often causes long-term equipment downtime and affects the normal operation of highways; on the other hand, although periodic inspections can detect some potential problems, they cannot achieve real-time monitoring of equipment status and early warning of faults, resulting in untimely fault handling and high maintenance costs.

[0004] In addition, although there are already some equipment monitoring systems in the prior art, most of them have the following deficiencies: one is the independent monitoring of single equipment, lacking cooperation between systems; the second is the simple alarm mechanism, mostly based on threshold judgment, unable to perform trend analysis and prediction; the third is the lack of intelligent decision support for fault handling, with low processing efficiency; the fourth is the lack of a closed-loop feedback mechanism, unable to evaluate the treatment effect and adjust in time.

[0005] Therefore, there is an urgent need for a system that can perform fault early warning and intelligent regulation to improve the management and maintenance level of highway electromechanical equipment. Summary of the Invention

[0006] The object of the present invention is to provide a fault early warning and intelligent regulation system for highway electromechanical equipment, which realizes early warning of faults by real-time monitoring the working state of electromechanical equipment and constructing a fault development trend model; and accurately diagnoses the fault type and cause through multi-source information fusion analysis; further generates a processing strategy based on an intelligent decision algorithm, and at the same time establishes a closed-loop feedback mechanism to ensure the effectiveness of fault handling, thereby improving the operation reliability and maintenance efficiency of highway electromechanical equipment.

[0007] The present invention proposes a fault early warning and intelligent regulation system for highway electromechanical equipment, including: A fault feature extraction and trend modeling module, used for: Collecting working state data from the sensor nodes of electromechanical equipment; Extracting multi-dimensional features of the working state data; Performing clustering analysis on the working state data to construct a fault development trend and evolution law model; A multi-source information fusion and fault diagnosis module, connected to the fault feature extraction and trend modeling module, is used for: Receiving the fault development trend and evolution law model sent by the fault feature extraction and trend modeling module; Collecting the response signals, detection information, and historical alarm information of the electromechanical equipment; Performing differential processing on the response signals to obtain the measured values between the electromechanical equipment; Generating analysis and evaluation data on the working state of the electromechanical equipment based on the measured values and the fault development trend and evolution law model; A fault handling strategy generation module, connected to the multi-source information fusion and fault diagnosis module, is used for: Receiving the analysis and evaluation data sent by the multi-source information fusion and fault diagnosis module; Generating fault handling instructions for the electromechanical equipment based on the analysis and evaluation data; A closed-loop feedback and emergency handling module, connected to the fault handling strategy generation module, is used for: Receiving the fault handling instructions sent by the fault handling strategy generation module; Sending the fault handling instructions to the corresponding electromechanical equipment; Monitoring the response of the electromechanical equipment to the fault handling instructions; When the electromechanical equipment fails to respond to the fault handling instructions, implementing emergency handling measures.

[0008] Preferably, the fault feature extraction and trend modeling module includes: A data preprocessing unit, used for: Cleaning and smoothing the working state data; Detecting and processing outliers in the working state data; A multi-dimensional feature extraction unit, connected to the data preprocessing unit, is used for: Extracting statistical features from the time domain; Extracting energy distribution features from the frequency domain; Extracting the operation mode conversion features of the electromechanical equipment; A pattern recognition and clustering unit, connected to the multi-dimensional feature extraction unit, is used for: Performing clustering analysis on the multi-dimensional features; Constructing a fault mode library according to the clustering results; A trend prediction and evolution rule unit, connected to the pattern recognition and clustering unit, is used for: Analyzing the development path of historical faults based on the fault mode library; Identifying fault triggering conditions and accelerating factors; Build a model for the fault development trend and evolution law; Send the fault development trend and evolution law model to the multi-source information fusion and fault diagnosis module.

[0009] Preferably, the multi-source information fusion and fault diagnosis module includes: An information source management unit for: Receiving the sensor data, self-check data, and historical alarm data of the electromechanical equipment respectively; Performing standardization processing on the sensor data, self-check data, and historical alarm data; A differential signal processing unit connected to the information source management unit for: Establishing a reference model based on historical normal operation data; Calculating the difference between the actual signal and the reference signal of the electromechanical equipment; Subtracting the signals between electromechanical equipment from each other to eliminate common interference; An information fusion engine connected to the differential signal processing unit for: Performing hierarchical fusion processing on data from different information sources; Establishing a spatial topology relationship model between devices; Analyzing the time series correlation of device state changes; A fault diagnosis unit connected to the information fusion engine for: Identifying the abnormal state of the electromechanical equipment based on the fusion processing result; Determining the fault type and severity; Generating analysis and evaluation data; Sending the analysis and evaluation data to the fault handling strategy generation module.

[0010] Preferably, the fault handling strategy generation module includes: A device state evaluation unit for: Receiving the analysis and evaluation data; Dividing the state of the electromechanical equipment into four levels: normal, sub-healthy, minor fault, and severe fault; Constructing a health score for the electromechanical equipment; A decision optimization unit connected to the device state evaluation unit for: Determining the weight distribution of the processing scheme based on multi-dimensional objectives such as safety, reliability, economy, and timeliness; Constructing a decision tree for processing measures to cover different fault scenarios; An instruction generation unit connected to the decision optimization unit for: Selecting a standard instruction template according to the decision result; Optimize and adjust the instruction parameters based on the status and fault characteristics of the electromechanical equipment; Generate fault handling instructions; Send the fault handling instructions to the closed-loop feedback and emergency handling module; The maintenance plan coordination unit, connected to the instruction generation unit, is used for: Formulate short-term, medium-term, and long-term maintenance plans based on the fault handling instructions; Optimize the resource allocation of maintenance tasks.

[0011] Preferably, the closed-loop feedback and emergency handling module includes: The instruction execution monitoring unit, used for: Receive the fault handling instructions; Send the fault handling instructions to the corresponding electromechanical equipment; Record the reference state of the electromechanical equipment before the instruction execution; Real-time monitor the changes in key parameters of the electromechanical equipment; Evaluate the progress and effect of the instruction execution; The response effect evaluation unit, connected to the instruction execution monitoring unit, is used for: Compare the state changes before and after the instruction execution; Judge the response of the electromechanical equipment to the instruction; Calculate the processing effect score; The feedback processing unit, connected to the response effect evaluation unit, is used for: Feedback the execution result to the fault handling strategy generation module; Update the fault handling knowledge base; The emergency handling unit, connected to the response effect evaluation unit, is used for: When the response effect evaluation unit determines that the electromechanical equipment fails to respond, evaluate the emergency level; Select the corresponding emergency plan according to the emergency level; Execute the emergency handling measures; Monitor the execution effect of the emergency measures.

[0012] Preferably, the multi-dimensional feature extraction unit is also used for: Extract time series features such as temperature change rate, working current fluctuation, and vibration spectrum; Extract the equipment working cycle and load change pattern; Extract the correlation features and mutual influence features between equipment; Construct a multi-dimensional feature vector according to the extracted features.

[0013] Preferably, the hierarchical fusion method adopted by the information fusion engine includes: The first - layer fusion: Fuse the data of the same - type information sources; The second - layer fusion: Fuse the features of different - type information; The third - layer fusion: Comprehensively fuse the multi - dimensional judgment results; The information fusion engine also dynamically adjusts the reliability weights according to the historical accuracy of the information sources.

[0014] Preferably, the decision - making constraints of the decision - making optimization unit include: Safety constraint: Ensure that the processing measures do not cause new safety risks; Resource constraint: Consider the availability of maintenance resources and scheduling limitations; Performance constraint: Guarantee the overall functions and performance requirements of the system; Cost constraint: Balance the maintenance cost and the downtime loss.

[0015] Preferably, the emergency - handling unit classifies the emergency situations into four levels: Prompt level: The equipment has minor abnormalities but does not affect normal operation; Early - warning level: The performance of the equipment has declined but is still within an acceptable range; Warning level: The equipment has functional failures and needs to be processed in a timely manner; Emergency level: The equipment has serious failures, which may lead to system crashes or safety accidents.

[0016] Preferably, the degraded - operation control of the emergency - handling unit includes: Define the core functions and non - core functions of the system; Implement selective function degradation to guarantee the core functions; Adjust the performance parameters to reduce the system load; Establish temporary alternative solutions to maintain basic services; Formulate system recovery plans and procedures.

[0017] The present invention has the following beneficial effects: 1. Realize the transformation from passive maintenance to active prevention. By predicting the fault trend and modeling the evolution law, potential risks can be identified at the budding stage of the fault, and preventive measures can be taken in advance, significantly reducing the sudden failures of equipment.

[0018] 2. Improve the accuracy of fault diagnosis. By adopting the multi - source heterogeneous information fusion technology, comprehensively analyze various types and multiple - source information, significantly improve the comprehensiveness and accuracy of fault diagnosis, and reduce the misjudgment rate and missed - judgment rate.

[0019] 3. Optimize the fault - handling decision - making. Based on the multi - objective optimization decision - making mechanism, comprehensively consider multi - dimensional objectives such as safety, reliability, economy, and timeliness, generate the optimal handling strategy, and improve the handling efficiency.

[0020] 4. Enhance system reliability. Through the closed-loop feedback and emergency handling mechanism, the execution effect of the processing instructions is monitored in real time. In case of instruction failure, emergency measures are quickly initiated to avoid the expansion of faults and ensure the safe and stable operation of the system.

[0021] 5. Reduce maintenance costs and extend equipment life. Preventive maintenance reduces emergency repair costs, reduces unplanned downtime, extends the service life of equipment, and significantly reduces the total cost of ownership. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is the overall architecture diagram of the highway electromechanical equipment fault warning and intelligent regulation system of the present invention; Figure 2 is the structural diagram of the fault feature extraction and trend modeling module of the present invention; Figure 3 is the structural diagram of the multi-source information fusion and fault diagnosis module of the present invention; Figure 4 is the structural diagram of the fault handling strategy generation module of the present invention; Figure 5 is the structural diagram of the closed-loop feedback and emergency handling module of the present invention; Figure 6 is the flow chart of clustering analysis and fault mode construction of the present invention; Figure 7 is the flow chart of multi-source information fusion processing of the present invention; Figure 8 is the flow chart of fault handling instruction generation of the present invention; Figure 9 is the emergency handling flow chart of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] Please refer to the attached Figures 1-9 , and the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0024] As Figure 1 shown, the highway electromechanical equipment fault warning and intelligent regulation system provided by the present invention includes a fault feature extraction and trend modeling module 1, a multi-source information fusion and fault diagnosis module 2, a fault handling strategy generation module 3, and a closed-loop feedback and emergency handling module 4.

[0025] The fault feature extraction and trend modeling module 1 is used to collect the working state data from the sensor nodes of the electromechanical equipment, extract the multi-dimensional features of the working state data, and perform clustering analysis on the working state data to construct a fault development trend and evolution law model.

[0026] The multi-source information fusion and fault diagnosis module 2 is connected to the fault feature extraction and trend modeling module 1, and is used to receive the fault development trend and evolution law model sent by the fault feature extraction and trend modeling module 1, collect the response signals, detection information and historical alarm information of the electromechanical equipment, perform differential processing on the response signals to obtain the measured values between the electromechanical equipment, and generate analysis and evaluation data on the working state of the electromechanical equipment based on the measured values and the fault development trend and evolution law model.

[0027] The fault handling strategy generation module 3 is connected to the multi-source information fusion and fault diagnosis module 2, and is used to receive the analysis and evaluation data sent by the multi-source information fusion and fault diagnosis module 2, and generate a fault handling instruction for the electromechanical equipment based on the analysis and evaluation data.

[0028] The closed-loop feedback and emergency handling module 4 is connected to the fault handling strategy generation module 3, and is used to receive the fault handling instruction sent by the fault handling strategy generation module 3, send the fault handling instruction to the corresponding electromechanical equipment, monitor the response of the electromechanical equipment to the fault handling instruction, and execute emergency handling measures when the electromechanical equipment fails to respond to the fault handling instruction.

[0029] In a preferred embodiment of the present invention, as Figure 2 shown, the fault feature extraction and trend modeling module 1 includes a data preprocessing unit 11, a multi-dimensional feature extraction unit 12, a pattern recognition and clustering unit 13, and a trend prediction and evolution rule unit 14.

[0030] The data preprocessing unit 11 is used to clean and smooth the working state data, and detect and process the outliers in the working state data. In practical applications, this unit uses the sliding window technology to detect outliers, and the window size can be dynamically adjusted according to the data characteristics, usually between 30 seconds and 300 seconds. Preferably, the three-sigma principle is used to screen out the outliers, that is, when the data point deviates from the mean by more than 3 times the standard deviation, it is determined as an outlier. For the detected outliers, they can be replaced by the previous value filling or linear interpolation method to ensure the continuity of the data. For example, for the carbon monoxide concentration sensor data in the highway tunnel, if the concentration value at a certain moment is detected as 85 ppm, while the concentrations at the previous and subsequent moments are 45 ppm and 47 ppm respectively, then this 85 ppm is very likely to be an outlier, and the system will use 46 ppm to replace it to ensure the accuracy of subsequent analysis.

[0031] The multi-dimensional feature extraction unit 12 is connected to the data preprocessing unit 11 and is used to extract statistical features from the time domain perspective, extract energy distribution features from the frequency domain perspective, and extract the operating mode conversion features of the electromechanical equipment. Specifically, the time domain features include statistics such as mean, variance, kurtosis, and skewness, which can accurately describe the distribution characteristics of the data; the frequency domain features are obtained through the fast Fourier transform (FFT) and include the main frequency, energy distribution, spectral entropy, etc., which can reflect the vibration characteristics of the equipment; while the operating mode conversion features focus on the conversion frequency and duration of the equipment between different working states, which helps to identify unstable working states. In the fault warning of highway lighting equipment, the system will extract features such as current fluctuation frequency, power factor change rate, temperature fluctuation mode, etc. For example, the power factor continuously being lower than 0.85 may indicate an increased risk of transformer failure.

[0032] The pattern recognition and clustering unit 13 is connected to the multi-dimensional feature extraction unit 12 and is used to perform clustering analysis on the multi-dimensional features and construct a fault mode library according to the clustering results. Preferably, this unit adopts an adaptive clustering mechanism and automatically selects a suitable clustering algorithm according to the data distribution characteristics. When the data shows an obvious hierarchical structure, the hierarchical clustering algorithm is adopted; when the data distribution is irregular, the density clustering algorithm is adopted. The number of clusters is automatically determined by the silhouette coefficient, usually between 3 and 7, to balance the clustering accuracy and computational efficiency. For highway toll collection equipment, typical fault clusters may include categories such as communication anomalies, card reading failures, printer failures, etc. The system establishes warning indicators by identifying similar fault patterns.

[0033] The core step of clustering analysis is to group similar fault patterns together, and its mathematical expression is: , where: is the feature vector between is the Euclidean distance, is the feature dimension, represents the th sample's th eigenvalue, represents the th sample's

[0034] Based on the distance calculation, the clustering process can be expressed as minimizing the following objective function: , where: is the clustering objective function value, used to evaluate the clustering quality, and the smaller the value, the better the clustering effect; is the number of clusters, determined according to the types and failure modes of highway electromechanical equipment, usually ranging from 3 to 7; is the th cluster, representing a set of equipment failures with similar characteristics; is the center of the th cluster, indicating the typical characteristics of this type of failure; represents the distance from the sample to the cluster center, reflecting the similarity degree between the sample and the typical failure mode. For example, for the highway information board system, abnormal display, communication interruption, and unstable power supply may be classified into different failure clusters.

[0035] The trend prediction and evolution rule unit 14 is connected to the pattern recognition and clustering unit 13, and is used to analyze the development path of historical failures, identify failure triggering conditions and acceleration factors, construct a failure development trend and evolution law model based on the failure mode library, and send the failure development trend and evolution law model to the multi-source information fusion and fault diagnosis module 2. In practical applications, this unit realizes the multi-temporal prediction of key parameters, including short-term (5 minutes to 1 hour), medium-term (1 to 24 hours), and long-term (1 to 7 days) predictions. For different prediction periods, appropriate prediction algorithms are selected: exponential smoothing method is mostly used for short-term prediction, autoregressive integrated moving average (ARIMA) model for medium-term prediction, and a combination of regression analysis and time series analysis methods for long-term prediction. For example, for the bandwidth utilization rate of highway communication equipment, the following prediction model can be established: , where: is the predicted value at time, representing the predicted value of the bandwidth utilization rate at a future time point, with the unit of percentage; is the prediction step size, indicating how far into the future to predict, with the unit of hour; is the historical observation value, representing the actual bandwidth utilization rate at a past time, with the unit of percentage; is the historical prediction error, representing the difference between the past predicted value and the actual value; is the model constant term, reflecting the basic level of the bandwidth utilization rate; is the autoregressive term coefficient, reflecting the influence degree of historical data on the future; is the moving average term coefficient, reflecting the correction degree of historical errors on future predictions; and are the orders of the autoregressive term and the moving average term respectively, controlling the range of historical data considered by the model. The usual value range is 1 - 5, and the optimal value is determined according to the data characteristics through information criteria (such as AIC or BIC). For example, for the fan system in a highway tunnel, the prediction model may consider the operation data of the past 3 hours (p = 3) and the prediction errors of the past 2 hours (q = 2) to predict the operation status of the next 2 hours (h = 2), so as to timely detect potential failure risks.

[0036] Fault evolution rule extraction is to analyze the evolution process of historical faults and identify the causal relationships and propagation paths between faults. For the highway power supply system, a typical fault cascading rule may be: voltage fluctuation (slight) → voltage instability (moderate) → power outage (severe). By capturing this evolution rule, the system can identify potential severe fault risks at an early stage. For example, when the voltage fluctuation frequency of a certain power transformer increases by 50% within 24 hours, the system will issue a warning and recommend preventive inspections to avoid subsequent possible power outage accidents.

[0037] In another preferred embodiment of the present invention, as Figure 3 shown, the multi-source information fusion and fault diagnosis module 2 includes an information source management unit 21, a differential signal processing unit 22, an information fusion engine 23, and a fault diagnosis unit 24.

[0038] The information source management unit 21 is used to receive the sensor data, self-check data, and historical alarm data of the electromechanical equipment respectively, and perform standardization processing on the sensor data, self-check data, and historical alarm data. This unit maintains an information source directory, including various sensors, equipment self-checks, operation logs, and maintenance records. To ensure data quality, a reliability rating (1 - 5 levels) is assigned to each information source, with level 5 indicating the most reliable. The standardization processing of data uses the interval normalization method to uniformly convert data with different dimensions to the [0, 1] interval: , where: is the normalized value, dimensionless, with a value range of [0, 1]; is the original value, and the unit is related to specific parameters, such as temperature (°C), voltage (V), current (A), etc.; and are the historical minimum and maximum values of this parameter respectively, and the unit is the same as the original value . For example, for the lighting system in a highway tunnel, the original data range of the light intensity sensor may be 50 - 2000 lux. Through normalization processing, it is converted to the standard range of 0 - 1, which is convenient for comprehensive analysis with other parameters (such as current, temperature).

[0039] The differential signal processing unit 22 is connected to the information source management unit 21 and is used to establish a reference model based on historical normal operation data, calculate the difference between the actual signal and the reference signal of the electromechanical device, and perform subtraction processing on the signals between electromechanical devices to eliminate common interference. The reference model is constructed based on the historical data of the device in the normal state, and a normal operation range is determined for each monitoring parameter. When the device operates under similar working conditions, the difference between the actual signal and the reference model can reflect potential abnormalities. The differential processing between devices is an effective method to eliminate common interference such as environmental factors by subtracting the signals of the same type of devices in the same time period from each other. For example, for two adjacent electronic information boards of the same model on a highway, if the difference in their working temperatures exceeds 10 °C, it may indicate that the heat dissipation of one of the devices is abnormal and needs to be checked.

[0040] The mathematical expression of differential signal processing is: , where: is the differential signal of device at time , and the unit is the same as the original signal; is the actual signal of device at time , such as temperature (°C), voltage (V), etc.; is the signal value of the reference model at time , and the unit is the same as , representing the expected value under normal working conditions. For example, for the server of the highway toll collection system, if the actual CPU temperature is 75 °C and the expected value of the reference model is 65 °C, the differential signal is 10 °C, exceeding the preset threshold of 8 °C, and the system will issue a temperature anomaly warning.

[0041] The differential between devices is expressed as: , where: is the signal difference between device and device at time , and the unit is the same as the original signal; is the actual signal of device at time ; is the actual signal of device at time . For the same model weather stations installed on a highway, if the difference in the wind speeds measured by the two devices exceeds 5 m / s and they are only 100 meters apart, this may indicate that the wind speed sensor of one of the devices is faulty.

[0042] The information fusion engine 23 is connected to the differential signal processing unit 22 and is used to perform hierarchical fusion processing on data from different information sources, establish a spatial topology relationship model between devices, and analyze the time-series correlation of device state changes. This engine adopts a hierarchical fusion architecture, which includes three levels: the first level is the fusion of data from the same type of information sources, such as the fusion of data from multiple temperature sensors; the second level is the feature-level fusion of different types of information, such as the comprehensive analysis of different parameters such as temperature, vibration, and current; the third level is the decision-level fusion, which synthesizes multi-dimensional judgment results to form a final decision. During the information fusion process, the weights are dynamically adjusted according to the historical accuracy of the information sources to ensure the reliability of the fusion results.

[0043] The information fusion based on the DS evidence theory can be expressed as: , where: is the degree of support for the hypothesis after fusion, dimensionless, and its value range is [0,1], representing the credibility of the judgment of a certain fault type; and are the degrees of support for the hypotheses and from two evidence sources respectively, dimensionless, and their value ranges are [0,1], representing the degree of support for the fault judgment by different sensors or detection methods; is the conflict coefficient, dimensionless, and its value range is [0,1], representing the degree of contradiction between two evidence sources.

[0044] To solve the problem of algorithm instability in the case of high conflict, a conflict adaptive processing mechanism is introduced. This mechanism defines a function , and its value changes according to the relationship between the conflict degree and the threshold . The specific form of the function is as follows: When , , When , , where: is the fusion support degree after conflict processing, dimensionless, and its value range is [0,1]; is the conflict threshold, dimensionless, usually set to 0.9, representing the critical value of the conflict degree; is the conflict reallocation coefficient, dimensionless, and its value range is [0,1], representing what proportion of the conflict is reallocated to the intersection hypothesis. When When the system adopts an improved fusion rule to avoid algorithm crashes caused by high conflicts. For multiple evidence sources, the pairwise fusion and iteration method is still used. In addition, for high-conflict situations, an evidence discount technique can be introduced: , where: is the evidence support degree after discounting, dimensionless, and its value range is [0, 1]; is the credibility coefficient of the th evidence source, dimensionless, and its value range is [0, 1], which is determined based on historical accuracy; is the support degree of a completely uncertain hypothesis, dimensionless, and its value range is [0, 1]; represents the universal set in the judgment framework.

[0045] The spatial topology relationship model describes the physical and logical connection relationships between devices, which helps to analyze the propagation path of faults. For example, for a highway monitoring system, there are clear dependencies between front-end cameras, transmission networks, and back-end storage. Abnormalities in front-end devices may affect back-end services. When a fiber optic communication line fails on a certain section, it may cause data transmission interruptions for all monitoring devices within that section. The system will quickly locate the fault source based on the topology relationship and avoid processing each device separately.

[0046] The fault diagnosis unit 24 is connected to the information fusion engine 23 and is used to identify the abnormal state of the electromechanical device based on the fusion processing result, determine the fault type and severity, generate analysis and evaluation data, and send the analysis and evaluation data to the fault handling strategy generation module 3. This unit realizes multi-level fault identification, including three granularities: device level, subsystem level, and system level. The fault types are divided into functional faults (the device function is completely lost), performance faults (the device function degrades), intermittent faults (the fault appears intermittently), and persistent faults (the fault persists). The fault severity is divided into 15 levels, where level 1 is a minor abnormality and level 5 is a serious fault that endangers system safety.

[0047] Fault diagnosis adopts a probability inference method based on a Bayesian network: , where: is the posterior probability of the fault type under the given evidence , dimensionless, and its value range is [0, 1], indicating the possibility of judging a certain fault type based on the currently observed abnormal phenomena; is the evidence observed under the condition of the fault type The likelihood probability, dimensionless, with a value range of [0, 1], represents the possibility of observing a specific abnormal phenomenon when a certain fault exists; is the fault type The prior probability, dimensionless, with a value range of [0, 1], represents the probability of a fault occurring estimated based on historical data and expert experience. For example, for a highway vehicle detector, if the phenomenon of no data output is observed, the system will calculate the probability that this may be caused by a power failure (prior probability 0.2), "communication interruption" (prior probability 0.5), or sensor damage (prior probability 0.3), and select the cause with the highest probability as the diagnostic result.

[0048] In another preferred embodiment of the present invention, as Figure 4 shown, the fault handling strategy generation module 3 includes a device status evaluation unit 31, a decision optimization unit 32, an instruction generation unit 33, and a maintenance plan coordination unit 34.

[0049] The device status evaluation unit 31 is used to receive the analysis and evaluation data, classify the electromechanical device status into four levels: normal, sub-healthy, minor fault, and serious fault, and construct an electromechanical device health score. Among them, the normal state means that all parameters of the device are within the normal range; the sub-healthy state means that some parameters show slight abnormalities but do not affect the function; the minor fault means that some functions are affected but the device can still operate; the serious fault means that the key function fails or there is a safety risk. The electromechanical device health score adopts a quantization standard of 0 - 100, where 90 - 100 points represent the normal state, 70 - 89 points represent the sub-healthy state, 40 - 69 points represent the minor fault, and 0 - 39 points represent the serious fault.

[0050] The formula for optimizing the health score is: , where: is the health score, dimensionless, with a value range of [0, 100], and the higher the value, the better the device state; is the weight of the th parameter, dimensionless, satisfying , reflecting the influence degree of the parameter on the device health state; is the normalized current value of the th parameter, dimensionless, with a value range of [0, 1]; is the normalized normal reference value, dimensionless, with a value range of [0, 1]; and are the normalized maximum allowable value and minimum allowable value of this parameter respectively, dimensionless, with a value range of [0, 1]. The parameter normalization process adopts the following method: , Wherein: is the current value of the th parameter, and the unit is related to the specific parameter; and are respectively the allowable minimum and maximum values of the parameter, and the unit is the same as that of . By introducing a normalization processing step, it is ensured that parameters with different dimensions (such as temperature in °C, voltage in V, vibration in mm / s, etc.) are in the same standard interval [0,1] when calculating the health score, avoiding scoring deviations caused by dimensional differences.

[0051] The device status evaluation unit 31 also estimates the remaining service life of the device, comprehensively considering the current state of the device, historical operation data, and workload: The estimation formula for the remaining service life of the device is optimized as: , Wherein: is the estimated remaining service life, with the unit of hours or days; is the basic expected life, and the unit is the same as that of , representing the expected remaining life of the device under ideal conditions; is the health score, dimensionless, and the value range is [0,100] is the comprehensive influence factor, dimensionless, and usually the value range is [0.5,1.2], reflecting the comprehensive influence of workload and environmental conditions on the life. The calculation method of the comprehensive influence factor is: , Wherein: and are respectively the weight coefficients of the load influence and the environmental influence, dimensionless, and satisfy , reflecting the relative importance of different factors on the life; is the load adjustment factor, dimensionless, and usually the value range is [0.6,1.2], reflecting the influence of workload on the life; is the environmental adjustment factor, dimensionless, and usually the value range is [0.7,1.1], reflecting the influence of environmental conditions on the life.

[0052] The decision optimization unit 32 is connected to the device status evaluation unit 31, and is used to determine the weight distribution of the treatment plan based on multi-dimensional objectives such as safety, reliability, economy, and timeliness, and construct a decision tree for treatment measures to cover different fault scenarios. This unit considers various decision constraints, including safety constraints (ensuring that the treatment measures do not cause new safety risks), resource constraints (considering the availability of maintenance resources), performance constraints (ensuring the overall function and performance of the system), and cost constraints (balancing the maintenance cost and the downtime loss).

[0053] The mathematical model for multi-objective decision optimization is as follows: , , , , where: is the objective function vector, which contains optimization objectives, such as safety, reliability, economy, timeliness, etc.; is the th objective function, and its meaning is determined by the specific objective; and are the inequality and equality constraints respectively, representing various resource limitations and system requirements; and are the numbers of the inequality and equality constraints respectively; is the decision variable, representing possible treatment plans; is the feasible region of the decision variable, representing the set of all feasible treatment plans. For example, for the fire protection system of a highway tunnel, the objective functions may include maximizing system safety, minimizing maintenance costs, and minimizing maintenance time, while the constraint conditions include maintaining basic fire protection functions during maintenance and not exceeding the budget limit, etc.

[0054] In practical applications, the weighted sum method is used to transform the multi-objective problem into a single-objective problem: , where: is the weight of the th objective, dimensionless, satisfying = 1, reflecting the relative importance of the objective. For core highway equipment such as the central control system, the weights of safety and reliability are usually set to 0.4 and 0.3, while the weights of economy and timeliness can be set to 0.15. When dealing with faults in the tunnel fire alarm system, the safety weight may be increased to 0.6, while for non-critical lighting systems, the economy weight may be increased to 0.3.

[0055] The instruction generation unit 33 is connected to the decision optimization unit 32 and is used to select a standard instruction template according to the decision result, optimize and adjust the instruction parameters based on the state and fault characteristics of the electromechanical equipment, generate a fault handling instruction, and send the fault handling instruction to the closed-loop feedback and emergency handling module 4. This unit maintains a standardized instruction template library, including operation types such as parameter adjustment, restart, switching, and degraded operation. For specific faults, the system generates customized handling instructions through template combination and parameter optimization. For example, for the problem of blurred images in highway surveillance cameras, the system may generate sequential instructions including cleaning the lens, adjusting the focus parameters, and checking the transmission line, and set detailed execution standards for each step.

[0056] To ensure the consistency of the fault judgment criteria, a mapping relationship between the parameter threshold and the health score is established: , where: is the health threshold for the -level fault, dimensionless, and the value range is [0, 100]; is the threshold for the th parameter corresponding to the -level fault, and the unit is related to the specific parameter; is the normal reference value, and the unit is the same as ; and are the allowable maximum and minimum values of this parameter respectively, and the unit is the same as ;

[0057] For example, for the brightness of lighting equipment, it is set that a 10% - 20% decrease in brightness is mapped to a health score of 70 - 89 (sub - healthy state); a 21% - 30% decrease in brightness is mapped to a health score of 40 - 69 (minor fault state); a decrease in brightness > 30% is mapped to a health score < 40 (severe fault state).

[0058] By establishing a consistent mapping relationship between the parameter threshold and the health score, it is ensured that there will be no contradiction between single - parameter judgment and comprehensive health judgment, and the consistency and rationality of fault handling decisions are improved. For a 10% - 20% decrease in brightness, usually increasing the drive current by 10% is selected; while for a decrease in brightness exceeding 30%, it is recommended to replace the lamp.

[0059] The maintenance plan coordination unit 34 is connected to the instruction generation unit 33 and is used to formulate short-term, medium-term, and long-term maintenance plans based on the fault handling instructions and optimize the resource allocation of maintenance tasks. The short-term plan covers emergency maintenance tasks within 24 hours, and the medium-term and long-term plans cover preventive maintenance tasks in the next 7 days. This unit evaluates the availability of maintenance personnel, spare parts, and tools, and optimizes task sequencing and resource allocation. For example, when the system detects that multiple electronic displays on the highway need to be maintained, it will comprehensively rank them according to the importance of the equipment, the severity of the fault, the geographical location, and the availability of maintenance personnel, ensuring that the equipment in key areas is given priority for maintenance while minimizing the travel distance of the maintenance personnel and improving the operation efficiency.

[0060] In a further preferred embodiment of the present invention, as Figure 5 shown, the closed-loop feedback and emergency handling module 4 includes an instruction execution monitoring unit 41, a response effect evaluation unit 42, a feedback processing unit 43, and an emergency handling unit 44.

[0061] The instruction execution monitoring unit 41 is used to receive the fault handling instructions, send the fault handling instructions to the corresponding electromechanical equipment, record the reference state of the electromechanical equipment before the instruction execution, monitor the changes of the key parameters of the electromechanical equipment in real time, and evaluate the progress and effect of the instruction execution. Before sending the instruction, this unit establishes a record of the reference state before execution as a reference for subsequent effect evaluation. During the instruction execution process, the system monitors the changes of key parameters in real time and sets multiple monitoring thresholds (warning / warning / danger). For example, for the voltage adjustment instruction of highway power supply equipment, the warning threshold can be set to ±5% of the nominal value, the warning threshold to ±8%, and the danger threshold to ±10%. When the real-time voltage value reaches 266V after the voltage adjustment instruction is executed, exceeding 20% of the 220V nominal value, a danger-level alarm will be immediately triggered, and the system will interrupt the current instruction execution and initiate emergency measures.

[0062] The response effect evaluation unit 42 is connected to the instruction execution monitoring unit 41 and is used to compare the state changes before and after the instruction execution, judge the response of the electromechanical equipment to the instruction, and calculate the processing effect score. This unit calculates the execution effect through the following formula: , where: is the effect score, dimensionless, with a value range of [0, 100], and the higher the value, the better the effect; is the weight of the th parameter, dimensionless, satisfying , reflecting the importance of the parameter to the effect evaluation; and are the parameter values before and after execution respectively, and the unit is related to the specific parameter; is the target parameter value, and the unit is related to and Same. When , it is determined that the response fails; It is a partial response; It is a valid response.

[0063] For example, for the bandwidth optimization instruction of highway communication equipment, the goal is to reduce the utilization rate from 95% to 70%. After execution, it actually drops to 75%. Then the effect score is: , At this time, it is determined as a valid response, but there is still room for optimization. The system may record this result as a reference for handling similar situations in the future.

[0064] The feedback processing unit 43 is connected to the response effect evaluation unit 42, and is used to feedback the execution result to the fault handling strategy generation module 3 to update the fault handling knowledge base. This unit is responsible for the systematic recording and analysis of the execution results, including successful cases and failed cases, to form a closed loop of continuous optimization. For successful cases, the system records the fault characteristics, handling measures, and effect scores as references for future similar faults; for failed cases, it analyzes the reasons for invalidity to avoid repeating mistakes. For example, when the system finds that for a certain type of tunnel lighting equipment, the method of increasing the drive current when the lamp brightness drops by 15% has a success rate as high as 95%, while for another type, the success rate of this method is only 60%, the system will automatically adjust the handling strategy to customize the optimal solution for different types of equipment.

[0065] The emergency handling unit 44 is connected to the response effect evaluation unit 42, and is used to evaluate the emergency level when the response effect evaluation unit 42 determines that the electromechanical equipment response fails, select the corresponding emergency plan according to the emergency level, execute the emergency handling measures, and monitor the execution effect of the emergency measures. This unit classifies the emergency situations into four levels: prompt level, warning level, alarm level, and emergency level. The prompt level means that the equipment has minor abnormalities but does not affect normal operation, such as slight fluctuations in some parameters; the warning level means that the equipment performance has declined but is still within an acceptable range, such as an increase in communication delay but not interruption; the alarm level means that the equipment has a functional failure and needs to be processed in time, such as partial function failure; the emergency level means that the equipment has a serious failure that may lead to system collapse or safety accidents, such as a complete failure of the power supply system.

[0066] For emergencies of different levels, the system takes corresponding handling measures. For example, for a communication equipment failure at the warning level, the system may activate a backup line; while for a power failure at the emergency level, it may immediately switch to a backup power supply and issue a network-wide alarm at the same time. In a typical scenario, when a carbon monoxide concentration sensor in a highway tunnel fails and the response handling measure fails, the system will evaluate it as the warning level, automatically activate the emergency plan, increase the operating frequency of the ventilation equipment, and at the same time notify the technical personnel to arrive at the scene for repair within 4 hours to ensure that the air quality in the tunnel is not affected.

[0067] In a more preferred embodiment of the present invention, the multi-dimensional feature extraction unit 12 is further configured to extract time-series features such as temperature change rate, operating current fluctuation, vibration spectrum, etc., extract the equipment operating cycle and load change pattern, extract the association features and mutual influence features between devices, and construct a multi-dimensional feature vector according to the extracted features.

[0068] For highway electromechanical equipment, the time-series feature is an important indicator reflecting the dynamic operating state of the equipment. The temperature change rate can be calculated by continuous temperature measurement: , where: is the temperature change rate, with the unit of °C / minute, indicating the degree of temperature change per unit time; and are the temperature values at times and respectively, with the unit of °C; is the sampling time interval, with the unit of minute. For a normally operating device, the temperature change rate should usually remain stable. A sudden rapid increase in temperature (such as / minute) may indicate overload or heat dissipation failure. For example, for a highway surveillance camera, in the normal operating state, the internal temperature change rate usually does not exceed 0.5 °C / minute. If it is suddenly detected that the temperature change rate reaches 2.5 °C / minute, the system will immediately issue an alarm for abnormal heat dissipation.

[0069] The operating current fluctuation feature is characterized by analyzing the mean value, variance, and peak factor of the current: , , where: is the current variance, with the unit of , reflecting the severity of the current fluctuation; is the current value at the th sampling point, with the unit of A; ; is the number of sampling points, usually 30 - 60 points; is the peak factor, dimensionless, representing the ratio of the maximum current to the effective value; is the maximum current value, with the unit of A. An abnormal increase in the peak factor (such as ) usually indicates that the device has an impact load or a short - circuit fault. For highway electronic display screens, the peak factor is usually between 2.0 - 2.8 during normal operation. If the detected peak factor reaches 4.2, it may indicate a short - circuit risk in the drive circuit.

[0070] The correlation characteristics between devices include signal transmission relationships, functional dependence relationships, etc. between upstream and downstream devices. For example, in a highway monitoring system, there is a clear data flow relationship among the front - end cameras, transmission networks, and back - end storage, which can be quantified by the correlation coefficient: , where: is the correlation coefficient between devices X and Y, dimensionless, with a value range of [- 1,1]. 1 represents a perfect positive correlation, - 1 represents a perfect negative correlation, and 0 represents no correlation; is the covariance, representing the degree of association of the change trends of two variables; and are the standard deviations of X and Y respectively; represents the expected value; and are the means of X and Y respectively. A sudden decrease in the correlation coefficient may indicate abnormal communication between devices. For example, in a highway vehicle detection system, the data of adjacent detectors usually has a high correlation (such as ). If the correlation coefficient suddenly drops to 0.3, it may indicate that one of the detectors is malfunctioning.

[0071] The construction of the multi - dimensional feature vector combines the above - mentioned various features to form a unified feature space: , where: is the feature vector, with the dimension of is the th feature component, and its meaning and unit are determined by the specific feature. For complex devices, the feature dimension may reach dozens or even hundreds. To improve the calculation efficiency, dimensionality reduction processing is usually carried out by methods such as principal component analysis (PCA).

[0072] When the multi - dimensional feature extraction unit performs PCA dimensionality reduction, a phased weight - assignment strategy is adopted: , where: is the weighted feature vector, with the dimension of is the diagonal weight matrix, with the dimension of , and the diagonal elements Represents the weight of the th original feature; is the original feature vector, with a dimension of . The PCA dimensionality reduction process is expressed as: , where: is the principal component vector after dimensionality reduction, with a dimension of ( is the feature vector matrix, with a dimension of , and each column corresponds to the direction of a principal component.

[0073] To maintain the weight consistency before and after dimensionality reduction, a feature subspace grouping strategy is adopted: 1. Features with similar dimensions are divided into different groups, such as the temperature group, vibration group, current group, etc.; 2. PCA dimensionality reduction is performed separately within each group; 3. Weights are assigned to the dimensionality-reduced subspaces instead of the original features; the subspace weight assignment is expressed as: , where: is the final feature vector; is the principal component vector of the th subspace; is the weight of the th subspace, dimensionless, satisfying , reflecting the importance of different feature groups; is the number of feature subspaces.

[0074] This optimization not only retains the introduction of expert knowledge (reflected by subspace weights) but also utilizes the dimensionality reduction advantage of PCA.

[0075] In another preferred embodiment of the present invention, the hierarchical fusion method adopted by the information fusion engine 23 includes: First-layer fusion: fusing data from the same type of information sources; Second-layer fusion: fusing features of different types of information; Third-layer fusion: comprehensively fusing multi-dimensional judgment results; The information fusion engine 23 also dynamically adjusts the reliability weights according to the historical accuracy of the information sources.

[0076] The first-layer fusion mainly processes data from the same type of sensors, such as the readings of multiple temperature sensors. The weighted average method is used for fusion: , where: is the fused signal value, with the same unit as the original signal; is the signal value of the th sensor, with the same unit as ; is the weight coefficient, dimensionless, and satisfies , reflecting the reliability and importance of each sensor; is the number of sensors. The weight coefficient can be determined according to the historical reliability of the sensors. For example, the weight of a sensor with a reliability score of 5 can be set to 0.5, while the weight of a sensor with a reliability score of 3 can be set to 0.3. In the highway traffic flow monitoring system, when multiple vehicle detectors monitor the same section, the system will automatically assign weights according to the historical accuracy of each detector to ensure the accuracy of the fusion result.

[0077] The second layer fuses different types of parameter information, such as temperature, vibration, current, etc. Feature-level fusion is adopted at this level to form a unified feature space by combining different parameters: , where: is the fused feature vector; is the feature vector of the th type of parameter; is the weight coefficient, dimensionless, reflecting the importance of different parameter types; is the number of parameter types. The weights of different types of parameters are determined according to their contribution to fault diagnosis. For example, for the fault diagnosis of highway electric doors, the weight of the motor current feature may be 0.4, the weight of the vibration feature is 0.3, the weight of the temperature feature is 0.2, and the weight of the sound feature is 0.1, comprehensively forming a comprehensive assessment of the fault state.

[0078] The third layer of fusion is decision-level fusion, which synthesizes multiple diagnostic results to form a final judgment: , where: is the final decision result, which may be a judgment on the fault type, fault location, or fault severity; is the result of the th diagnostic method; is the weight coefficient, dimensionless, reflecting the reliability and applicability of each diagnostic method; is the number of diagnostic methods. In practical applications, methods such as voting, DS evidence theory, or fuzzy inference can be used to achieve decision fusion. For example, for the failure of the carbon monoxide monitoring system in the highway tunnel, the system may simultaneously use a rule-based diagnostic method (weight 0.3), a model-based diagnostic method (weight 0.4), and a historical case-based diagnostic method (weight 0.3), and synthesize the judgment results of each method to obtain the final diagnosis.

[0079] The dynamic adjustment of the information source reliability weight is based on the historical accuracy evaluation: , Wherein: is the information source at time weight, dimensionless; is the accuracy rate in the recent period, dimensionless, and the value range is [0,1]; is the weight at the previous moment, dimensionless; is the update coefficient, dimensionless, usually taking a value of 0.2 - 0.3, indicating the influence degree of the new accuracy rate. This dynamic adjustment mechanism ensures that the system can adaptively optimize the information fusion process. For example, for a certain camera in the highway video surveillance system, if the abnormal detection accuracy rate in the past 24 hours is only 65% (i.e., 35% is false alarm), the system will automatically reduce the weight of this information source to reduce the impact of false alarms on the overall judgment.

[0080] In a further preferred embodiment of the present invention, the decision-making constraints of the decision optimization unit 32 include: safety constraint: ensuring that the processing measures do not cause new safety risks; resource constraint: considering the availability of maintenance resources and scheduling limitations; performance constraint: guaranteeing the overall function and performance requirements of the system; cost constraint: balancing the maintenance cost and the downtime loss.

[0081] The decision optimization process needs to consider multiple constraint conditions simultaneously to ensure that the generated processing plan is both effective and feasible. The safety constraint is the primary consideration factor to ensure that the processing measures do not cause new safety problems. For example, for the ventilation equipment in the highway tunnel, even if a fault is found, it cannot be completely shut down for maintenance during the peak period, but needs to be maintained in zones while ensuring basic ventilation.

[0082] The safety constraint can be expressed as: , Wherein: is the safety constraint function, indicating the constraint conditions of the decision-making scheme in terms of safety; is the decision-making scheme safety risk value, dimensionless, usually quantified in the range of [0,1]; is the maximum acceptable risk threshold, dimensionless, indicating the maximum risk level that the system can accept. For critical equipment, is usually set very low, such as 0.05 (5% risk probability). For example, when dealing with the failure of the highway traffic signal control system, if a certain maintenance plan may result in a signal error risk of 0.08, while the maximum acceptable risk set by the system is 0.05, then this plan will be rejected.

[0083] The resource constraint considers the availability of maintenance personnel, spare parts and tools: , Wherein: is the resource constraint function; is the decision-making plan The required amount of resources, which can be the number of maintenance personnel, man-hours, the number of spare parts, etc.; is the available amount of resources, and the unit is the same as that of Same. Resource constraints ensure that the generated maintenance plan can be executed under the existing conditions. For example, when lighting equipment failures occur simultaneously in multiple sections of a highway, the system will consider the actual available number of maintenance vehicles and technicians, and arrange a reasonable maintenance sequence and allocation.

[0084] Performance constraints focus on ensuring the overall function of the system: , Wherein: is the performance constraint function; is the decision-making plan The system performance indicators under, such as coverage rate, response time, etc.; is the minimum performance requirement, and the unit is the same as that of Same. For example, the highway monitoring system may require that the monitoring coverage rate is not less than 90% even when some equipment is under maintenance. During the maintenance of the tunnel lighting system, it may be required that the lighting brightness during maintenance is not less than 70% of the normal value to ensure driving safety.

[0085] Cost constraints balance maintenance costs and downtime losses: , Wherein: is the cost constraint function; is the maintenance cost, with the unit of yuan, including labor costs, material costs, equipment costs, etc.; is the downtime loss, with the unit of yuan, reflecting the economic loss caused by equipment downtime to the system operation; is the available budget, with the unit of yuan. In practical applications, the system will estimate the total cost of different maintenance strategies (direct maintenance costs plus indirect losses) and select the plan with the best economic benefits. For example, for a highway toll system failure, immediately replacing the entire set of equipment may have a high cost (such as 500,000 yuan) but a short downtime (2 hours), while component-by-component maintenance has a low cost (150,000 yuan) but a long downtime (10 hours). The system will calculate the optimal plan based on the current toll traffic and available budget.

[0086] In yet another preferred embodiment of the present invention, the emergency processing unit 44 classifies emergency situations into four levels: Prompt level: The equipment has minor abnormalities but does not affect normal operation; Warning level: The equipment performance deteriorates but is still within an acceptable range; Alarm level: The equipment has a functional failure and needs to be processed in a timely manner; Emergency level: The equipment has a serious failure, which may lead to system collapse or safety accidents.

[0087] The emergency level assessment is based on a comprehensive judgment of multi-dimensional indicators, including the scope of fault impact, development speed, and severity: , Among them: is the comprehensive score of the emergency level, dimensionless, usually in the range of [0, 10]; 、 and are the scores of the impact scope, development speed, and severity respectively, dimensionless, both in the range of [1, 10] points; 、 and are the weight coefficients, dimensionless, satisfying , usually set to 0.3, 0.3, and 0.4, reflecting the relative importance of each factor. According to the comprehensive score, the emergency level is divided into: is the prompt level, is the pre-warning level, is the warning level, is the emergency level.

[0088] For example, for the fault of abnormal display on the highway information board, if the system evaluates that the impact scope score is 4 points (only affecting a single device), the development speed is 3 points (stable and non-spreading), and the severity is 5 points (partial error in the displayed information), then: , This fault will be determined as the pre-warning level, and the system will arrange for maintenance within 24 hours and may temporarily adjust the display content to reduce the impact.

[0089] For different levels of emergency situations, the system adopts corresponding processing strategies: Prompt level (Level 1): The system records the abnormal situation, may adjust some parameters, but will not issue an alarm or interrupt normal operations. For example, when the image quality of a monitoring camera on a certain section slightly decreases, the system will record and arrange the next routine maintenance inspection, but will not immediately dispatch technical personnel.

[0090] Pre-warning level (Level 2): The system issues an internal pre-warning, may start some preventive measures, but will not affect the normal operation of the equipment. For example, when the temperature of the power supply equipment continues to rise but does not reach the dangerous value, the system may adjust the load distribution and arrange for technical personnel to check within 24 hours. For the backup power supply system in the highway tunnel with a decrease in efficiency, it may be adjusted to be tested once a week (normally tested once a month), and priority will be given to checking during the next maintenance.

[0091] Warning Level (Level 3): The system issues an external alarm and activates the emergency response plan, which may involve functional degradation or local alternative measures. For example, when a section of the toll collection system fails, the system will activate the backup toll collection mode and dispatch maintenance personnel urgently. When the vehicle detection system at the entrance of a highway tunnel fails, the system may enable a backup camera for temporary monitoring and require technicians to arrive at the scene within 4 hours.

[0092] Emergency Level (Level 4): The system triggers the highest-level alarm and immediately executes emergency measures, which may include equipment shutdown, system isolation, or comprehensive alternative solutions. For example, when the tunnel ventilation system fails completely, the system will immediately close the tunnel entrance, activate the emergency lighting, and mobilize all available resources for emergency repair. When a serious failure occurs in the highway traffic signal control system, traffic controllers may be immediately deployed to conduct on-site command, and at the same time, the backup control system will be activated.

[0093] The trigger condition for degraded operation is optimized to a multi-condition combination judgment mechanism. The trigger condition for Level 1 (minor degradation) is: , where: is the trigger status for Level 1 degradation, a boolean value, true indicates triggering degradation; is the system load, in percentage, indicating the utilization rate of system resources; is the first load threshold, in percentage, usually set to 85%; is the equipment health score, dimensionless, with a value range of [0, 100]; is the health threshold, dimensionless, usually set to 70 points; is the second load threshold, in percentage, usually set to 95%; is the key performance parameter (such as network latency, data packet loss rate, etc.), and the unit is related to the specific parameter; is the key parameter threshold, and the unit is the same as that of The optimized degradation trigger condition no longer solely depends on the system load, but comprehensively considers the equipment health status and key performance indicators, avoiding unnecessary degradation solely due to high load when the equipment is in good health, and improving the system availability.

[0094] In another preferred embodiment of the present invention, the degraded operation control of the emergency processing unit 44 includes: defining the core functions and non-core functions of the system; achieving selective function degradation to ensure core functions; adjusting performance parameters to reduce the system load; establishing a temporary alternative solution to maintain basic services; formulating a system recovery plan and steps.

[0095] Degraded operation is a key mechanism to ensure that the system can still provide basic services in case of failures. First of all, the system needs to clearly distinguish between core functions and non-core functions. For a highway monitoring system, real-time video monitoring is a core function, while historical data query and high-definition picture quality may be non-core functions; for a toll collection system, basic toll collection is the core, while ETC discount calculation and invoice printing may be non-core functions. The system can quickly determine which functions must be retained and which can be temporarily disabled in case of failures through a predefined function priority list.

[0096] Function selective degradation is a strategy to reduce non-core functions in a planned way according to the failure situation to ensure core functions. For example, when the storage system of a highway monitoring center fails, the system may reduce the video storage resolution from 1080p to 720p and extend the storage period from 30 days to 15 days to ensure continuous monitoring coverage. In case of limited communication bandwidth, the transmission of control signals and alarm information may be prioritized and non-critical data synchronization may be suspended.

[0097] Performance parameter adjustment is a method to reduce the system load by modifying working parameters. For example, for a highway lighting system, when the power supply is limited, the lighting brightness may be reduced to 80% of the safety standard to reduce power consumption; for a communication system, the data transmission priority may be reduced to ensure that key information is not affected. When a part of the tunnel ventilation system fails, the operating frequency of the fan may be adjusted to maintain the minimum ventilation effect while reducing the pressure on the faulty components.

[0098] Temporary alternative solutions are mechanisms to enable alternative means to maintain basic services when the original system cannot work properly. For example, when the electronic toll collection system fails, manual toll collection can be temporarily enabled; when the monitoring system fails, additional patrol personnel can be dispatched; when the message board display fails, road conditions can be announced through radio stations. In case of a failure of the carbon monoxide monitoring system in a tunnel, the ventilation frequency may be temporarily increased and the passage of large vehicles may be restricted to ensure air quality.

[0099] The system recovery plan includes function recovery steps, data synchronization methods and performance verification processes after troubleshooting. For example, for a highway toll collection system attacked by a network, the recovery plan may include steps such as system isolation, vulnerability repair, data recovery and security verification to ensure that the system can resume normal operation safely and reliably. For cases where hardware needs to be replaced, the plan will include detailed processes for equipment installation, parameter configuration, function testing and performance verification to ensure that the performance of the restored system meets the requirements.

[0100] The implementation of degraded operation control requires predefined multi-level degradation schemes and trigger conditions. For example, for a highway communication system, three-level degradation schemes can be defined: Level 1 (Minor Degradation) Trigger Conditions: 1. System load > 85% and Health Score < 70 (Sub-Health State) or 2. System load > 95% (Near Peak Load) or 3. Key parameters (such as network latency, data packet loss rate) exceed normal thresholds. This multi-condition combination trigger mechanism avoids unnecessary degradation solely due to high load when the device is in a healthy state, improves the system's availability and intelligence, and reduces the decline in service quality caused by mis-degradation.

[0101] In summary, the highway electromechanical equipment fault warning and intelligent regulation system provided by the present invention realizes the transformation from passive maintenance to active prevention through the collaborative work of four core modules: fault feature extraction and trend modeling, multi-source information fusion and fault diagnosis, fault handling strategy generation, and closed-loop feedback and emergency handling. It significantly improves the operation reliability and maintenance efficiency of highway electromechanical equipment and has important practical value.

[0102] Those skilled in the art should understand that the embodiments described in the specification and drawings of the present invention are exemplary, and the protection scope of the present invention is not limited to these embodiments. Any equivalent modifications or transformations made by those skilled in the art based on the disclosure of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A fault warning and intelligent regulation system for highway electromechanical equipment, characterized in that, Including: A fault feature extraction and trend modeling module, which is used for: Collecting the working state data from the sensor nodes of the electromechanical equipment; Extracting the multi-dimensional features of the working state data; Performing clustering analysis on the working state data to construct a fault development trend and evolution law model; A multi-source information fusion and fault diagnosis module, connected to the fault feature extraction and trend modeling module, which is used for: Receiving the fault development trend and evolution law model sent by the fault feature extraction and trend modeling module; Collecting the response signals, detection information and historical alarm information of the electromechanical equipment; Performing differential processing on the response signals to obtain the measured values between the electromechanical equipment; Generating analysis and evaluation data of the working state of the electromechanical equipment based on the measured values and the fault development trend and evolution law model; A fault handling strategy generation module, connected to the multi-source information fusion and fault diagnosis module, which is used for: Receiving the analysis and evaluation data sent by the multi-source information fusion and fault diagnosis module; Generating a fault handling instruction for the electromechanical equipment based on the analysis and evaluation data; A closed-loop feedback and emergency handling module, connected to the fault handling strategy generation module, which is used for: Receiving the fault handling instruction sent by the fault handling strategy generation module; Sending the fault handling instruction to the corresponding electromechanical equipment; Monitoring the response of the electromechanical equipment to the fault handling instruction; When the electromechanical equipment fails to respond to the fault handling instruction, performing emergency handling measures.

2. The system according to claim 1, wherein The fault feature extraction and trend modeling module includes: A data preprocessing unit, which is used for: Performing cleaning and smoothing processing on the working state data; Detecting and processing the outliers in the working state data; A multi-dimensional feature extraction unit, connected to the data preprocessing unit, which is used for: Extracting statistical features from the time domain perspective; Extracting energy distribution features from the frequency domain perspective; Extracting the operation mode conversion features of the electromechanical equipment; A pattern recognition and clustering unit, connected to the multi-dimensional feature extraction unit, which is used for: Performing clustering analysis on the multi-dimensional features; Constructing a fault mode library according to the clustering results; A trend prediction and evolution rule unit, connected to the pattern recognition and clustering unit, which is used for: Analyzing the development path of historical faults based on the fault mode library; Identifying the fault triggering conditions and acceleration factors; Constructing a fault development trend and evolution law model; Sending the fault development trend and evolution law model to the multi-source information fusion and fault diagnosis module.

3. The system according to claim 1, characterized in that, The multi-source information fusion and fault diagnosis module includes: An information source management unit, which is used for: Receiving the sensor data, self-check data and historical alarm data of the electromechanical equipment respectively; Performing standardization processing on the sensor data, self-check data and historical alarm data; A differential signal processing unit, connected to the information source management unit, which is used for: Establishing a reference model based on the historical normal operation data; Calculating the difference between the actual signal and the reference signal of the electromechanical equipment; Performing subtraction processing on the signals between the electromechanical equipment to eliminate the common interference; An information fusion engine, connected to the differential signal processing unit, which is used for: Performing hierarchical fusion processing on the data from different information sources; Establishing a spatial topology relationship model between the devices; Analyze the time - series correlation of equipment status changes; A fault diagnosis unit, connected to the information fusion engine, is used for: Identify the abnormal status of electromechanical equipment based on the fusion processing results; Determine the fault type and severity; Generate analysis and evaluation data; Send the analysis and evaluation data to the fault handling strategy generation module.

4. The system according to claim 1, characterized in that The fault handling strategy generation module includes: An equipment status evaluation unit, used for: Receive the analysis and evaluation data; Classify the status of electromechanical equipment into four levels: normal, sub - healthy, minor fault, and serious fault; Construct a health score for electromechanical equipment; A decision optimization unit, connected to the equipment status evaluation unit, is used for: Based on multi - dimensional goals of safety, reliability, economy, and timeliness, determine the weight allocation of treatment plans; Construct a decision tree for treatment measures, covering different fault scenarios; An instruction generation unit, connected to the decision optimization unit, is used for: Select a standard instruction template according to the decision result; Based on the status and fault characteristics of electromechanical equipment, optimize and adjust the instruction parameters; Generate fault handling instructions; Send the fault handling instructions to the closed - loop feedback and emergency handling module; A maintenance plan coordination unit, connected to the instruction generation unit, is used for: Based on the fault handling instructions, formulate short - term, medium - term, and long - term maintenance plans; Optimize the resource allocation of maintenance tasks.

5. The system according to claim 1, characterized in that, The closed - loop feedback and emergency handling module includes: An instruction execution monitoring unit, used for: Receive the fault handling instructions; Send the fault handling instructions to the corresponding electromechanical equipment; Record the baseline status of electromechanical equipment before instruction execution; Real - time monitor the changes in key parameters of electromechanical equipment; Evaluate the progress and effect of instruction execution; A response effect evaluation unit, connected to the instruction execution monitoring unit, is used for: Compare the status changes before and after instruction execution; Judge the response of electromechanical equipment to the instructions; Calculate the treatment effect score; A feedback processing unit, connected to the response effect evaluation unit, is used for: Feed back the execution result to the fault handling strategy generation module; Update the fault handling knowledge base; An emergency handling unit, connected to the response effect evaluation unit, is used for: When the response effect evaluation unit determines that the electromechanical equipment response fails, evaluate the emergency level; Select the corresponding emergency plan according to the emergency level; Execute emergency handling measures; Monitor the execution effect of emergency measures.

6. The system according to claim 2, wherein The multi - dimensional feature extraction unit is also used for: Extract the temperature change rate, working current fluctuation, and vibration spectrum time - series features; Extract the equipment working cycle and load change pattern; Extract the correlation features and mutual influence features between equipment; Construct a multi - dimensional feature vector according to the extracted features.

7. The system according to claim 3, characterized in that, The hierarchical fusion method adopted by the information fusion engine includes: The first - layer fusion: fuse the data of the same type of information sources; The second - layer fusion: fuse the features of different types of information; The third - layer fusion: comprehensively fuse the multi - dimensional judgment results; The information fusion engine also dynamically adjusts the reliability weights according to the historical accuracy of information sources.

8. The system according to claim 4, wherein The decision constraints of the decision optimization unit include: Safety constraint: Ensure that the treatment measures do not cause new safety risks; Resource constraint: Consider the availability of maintenance resources and scheduling limitations; Performance constraints: Ensure the overall system functions and performance requirements; Cost constraints: Balance maintenance costs and downtime losses.

9. The system according to claim 5, wherein The emergency handling unit classifies emergency situations into four levels: Prompt level: Slight anomalies occur in the equipment, but it does not affect normal operation; Warning level: The equipment performance degrades, but it is still within an acceptable range; Alarm level: Functional failures occur in the equipment and need to be handled promptly; Emergency level: Severe failures occur in the equipment, which may lead to system crashes or safety accidents.

10. The system according to claim 5, wherein The degraded operation control of the emergency handling unit includes: Define the core functions and non-core functions of the system; Implement selective function degradation to ensure core functions; Adjust performance parameters to reduce the system load; Establish temporary alternative solutions to maintain basic services; Formulate system recovery plans and procedures.

Images