Airport equipment health monitoring system

Abstract

The invention provides an airport equipment health monitoring system, and relates to the technical field of intelligent operation and maintenance. Comprises: a data acquisition module configured at a key node of a target device and collecting multi-source data of the device in real time; the data processing module is used for performing data integration on the multi-source data to obtain input data; the health assessment module is used for generating an equipment health index and a health degree by using the input data according to a machine learning model of multi-modal data fusion and identifying a health problem type; the health problem types comprise abnormal equipment working state, reduced equipment response speed, too high equipment working temperature, over-limit equipment load and insufficient equipment space margin; and the early warning response module is used for triggering a directional early warning signal according to the equipment health index and generating an operation and maintenance work order containing fault positioning information. According to the invention, the problems of insufficient detection precision, lagging response speed and poor environmental adaptability in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent operation and maintenance technology, and in particular to an airport equipment health monitoring system. Background Art

[0002] In modern airport operations, the health of various equipment (such as baggage handling systems, security equipment, and information display screens) directly impacts airport safety and efficiency. However, with the increasing variety and quantity of equipment, equipment operation and maintenance management faces serious challenges, especially in comprehensive performance testing. Current equipment monitoring methods often fail to efficiently and accurately assess the multi-dimensional health of equipment, particularly in the following areas:

[0003] Insufficient detection accuracy: Existing technologies are generally unable to provide comprehensive health assessments, resulting in untimely identification of equipment problems and the risk of fault propagation.

[0004] Slow response speed: There is often a delay in the feedback from the equipment to the monitoring platform, which brings challenges to operation and maintenance decision-making.

[0005] Load capacity monitoring: Lack of real-time monitoring of equipment under high load conditions affects fault warning.

[0006] Poor environmental adaptability: Environmental factors such as the equipment's operating temperature and humidity are not effectively monitored, resulting in performance degradation and increased risk of failure. Summary of the Invention

[0007] In order to overcome the deficiencies of the prior art, the present invention aims to provide an airport equipment health monitoring system.

[0008] To achieve the above object, the present invention provides the following solutions:

[0009] An airport equipment health monitoring system, comprising:

[0010] The data acquisition module is configured at the key nodes of the target device to collect multi-source data of the device in real time;

[0011] The data processing module is used to integrate the multi-source data to obtain input data;

[0012] The health assessment module is used to generate a device health index and health level using input data based on a machine learning model that integrates multimodal data, and to identify health issue types, including abnormal device operating status, reduced device response speed, excessive device operating temperature, excessive device load, and insufficient device space margin.

[0013] The early warning response module is used to trigger a directional early warning signal based on the equipment health index and generate an operation and maintenance work order containing fault location information.

[0014] Preferably, the multi-source data includes: operating parameters, instantaneous load parameters, environmental parameters and electrical status.

[0015] Preferably, the data acquisition module includes:

[0016] Node determination submodule, node identification submodule, sensor placement submodule;

[0017] The node determination submodule is used to determine the key nodes of the target device, the node identification submodule is used to identify the type of the key node and obtain an identification result, and the sensor placement submodule is used to place corresponding sensors at the key node positions according to the identification result.

[0018] Preferably, the data processing module includes:

[0019] Preprocessing submodule, first feature extraction submodule, second feature extraction submodule and fusion submodule;

[0020] The preprocessing submodule is used to perform time series alignment, outlier detection and outlier correction on the multi-source data to obtain preprocessed multi-source data, wherein the preprocessed multi-source data includes: preprocessed operating parameters, preprocessed instantaneous load parameters, preprocessed environmental parameters and preprocessed electrical states; the first feature extraction submodule is used to extract characteristic signals of the preprocessed operating parameters, preprocessed instantaneous load parameters and preprocessed environmental parameters to obtain a parameter feature signal set; the second feature submodule is used to use Clark transform and Park transform to extract voltage signals in the preprocessed electrical state to obtain direct-axis and quadrature-axis voltages; the fusion submodule is used to perform feature fusion on the parameter feature signal set and the direct-axis and quadrature-axis voltages to obtain input data.

[0021] Preferably, the expression of the parameter characteristic signal set is:

[0022] ;

[0023] in, is the parameter characteristic signal set, is the time series of the i-th operating parameter, is the mean of the time series of the operating parameters after time alignment, is the standard deviation of the time series of the operating parameters after time alignment, is the weight coefficient, is the nonlinear amplification coefficient, is the instantaneous load value, is the load safety threshold, is the jth environmental parameter, For environmental reference benchmarks, is the environmental coupling coefficient, n is the number of monitoring types of equipment operating parameters, and m is the number of monitoring types of environmental parameters.

[0024] Preferably, the expression of the input data is:

[0025] ;

[0026] in, is the weight matrix, and are the direct-axis / quadrature-axis voltages extracted by Clark transform and Park transform, respectively. is the electrical characteristic weight matrix, is the bias term, is a non-linear activation function.

[0027] Preferably, the health assessment module includes:

[0028] Dimensionality reduction submodule, model building submodule, health index calculation submodule, health assessment submodule and identification submodule;

[0029] The dimensionality reduction submodule is used to reduce the dimensionality of the input data to obtain the input data after dimensionality reduction. The model construction submodule is used to construct a health index calculation model based on the machine learning model. The health index calculation submodule is used to obtain the output of the health index calculation model and obtain a health index based on the output. The health assessment submodule is used to determine the current health level based on the health index, wherein the health level is divided into excellent, good, medium, and poor. The identification submodule is used to determine the type of health problem based on the health index.

[0030] Preferably, the expression of the health index calculation model is:

[0031] ;

[0032] in, is the input data after dimensionality reduction, For the The encoder output of the autoencoder, is the dynamic attention weight, is the output of the health index calculation model, is the Sigmoid function, K is the total number of autoencoder groups, and k is the index of the autoencoder.

[0033] Preferably, the identification submodule includes:

[0034] separation units, extraction units, distribution units, and classification units;

[0035] The classification unit is used to receive the input data after dimensionality reduction and split the input data after dimensionality reduction into three independent feature sets. The extraction unit is used to extract key features of the three independent feature sets to obtain key feature sets. The allocation unit is used to perform weight allocation on the key feature sets to obtain allocated feature sets. The classification unit is used to obtain the type of health problem based on the trained classifier and the allocated feature sets.

[0036] The present invention discloses the following technical effects:

[0037] The present invention provides an airport equipment health monitoring system, comprising: a data acquisition module configured to be located at key nodes of target equipment and collect multi-source data from the equipment in real time; a data processing module for integrating the multi-source data to obtain input data; a health assessment module for generating an equipment health index and health level using the input data and identifying health problem types based on a machine learning model that fused multimodal data. The health problem types include: abnormal equipment operating status, reduced equipment response speed, excessive equipment operating temperature, excessive equipment load, and insufficient equipment space margin; and an early warning response module for triggering a targeted early warning signal based on the equipment health index and generating an operation and maintenance work order containing fault location information. The present invention enhances the accuracy of health assessments by collecting multi-source data such as operating parameters, instantaneous load, environmental parameters, and electrical status in real time. A machine learning model is used to generate a health index, promptly identify potential faults, and prevent them from spreading. Real-time monitoring by the data acquisition module reduces feedback delays and ensures rapid information transmission. The early warning response module can trigger signals instantly, allowing operation and maintenance personnel to take prompt measures and improve response efficiency; the system dynamically monitors instantaneous load parameters to ensure real-time control of equipment under high load conditions, promptly identify load overload problems, and reduce the risk of failure; monitoring of environmental parameters enables timely evaluation and adjustment of equipment under adverse conditions, reducing the risk of performance degradation and ensuring that equipment operates in the optimal environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 A schematic diagram of the structure of an airport equipment health monitoring system provided by an embodiment of the present invention.

[0040] Description of reference numerals:

[0041] 1-Data acquisition module, 2-Data processing module, 3-Health assessment module, 4-Early warning response module. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] like Figure 1 As shown, the present invention provides an airport equipment health monitoring system, comprising:

[0045] Data acquisition module 1, configured at key nodes of the target device, to collect multi-source data of the device in real time;

[0046] The data processing module 2 is used to integrate the multi-source data to obtain input data;

[0047] The health assessment module 3 is used to generate a device health index and health level using input data based on a machine learning model fused with multimodal data, and to identify health problem types, including abnormal device operating status, reduced device response speed, excessive device operating temperature, excessive device load, and insufficient device space margin.

[0048] The early warning response module 4 is used to trigger a directional early warning signal according to the equipment health index and generate an operation and maintenance work order containing fault location information.

[0049] Furthermore, the multi-source data includes: operating parameters, instantaneous load parameters, environmental parameters and electrical status.

[0050] Furthermore, the data acquisition module 1 includes:

[0051] Node determination submodule, node identification submodule, sensor placement submodule;

[0052] The node determination submodule is used to determine the key nodes of the target device, the node identification submodule is used to identify the type of the key node and obtain an identification result, and the sensor placement submodule is used to place corresponding sensors at the key node positions according to the identification result.

[0053] Specifically, the node determination submodule has the following functions:

[0054] Target equipment analysis: Analyze the structure and working principle of the equipment to determine which parts are critical to the operation of the equipment.

[0055] Key node selection: Based on equipment specifications and actual operational requirements, key monitoring nodes are selected. These nodes typically include motors, transmission systems, control systems, and power modules.

[0056] Example nodes: rotor and stator of an electric motor;

[0057] Key oil circuit nodes of hydraulic system;

[0058] The heat transfer node of the cooling system.

[0059] The node identification submodule functions as follows:

[0060] Node type identification: Node type identification is performed using image recognition technology, sensor feedback, or equipment manuals.

[0061] Identification result output: Generates the type identification results of key nodes. For example:

[0062] Temperature sensor node;

[0063] Pressure sensor node;

[0064] Current sensor node.

[0065] Sensor placement submodule:

[0066] Sensor selection: Based on the node identification results, appropriate types of sensors are selected for deployment to ensure that the required multi-source data can be obtained.

[0067] Sensor types include:

[0068] Vibration sensor:

[0069] Type: accelerometer (such as MEMS accelerometer);

[0070] Function: Monitor the vibration status of the equipment and identify abnormal vibration patterns to indicate potential mechanical failures.

[0071] Torque sensor:

[0072] Type: electromagnetic torque sensor;

[0073] Function: Monitor the torque changes of the equipment during operation and determine whether the equipment load is normal.

[0074] Current sensor:

[0075] Type: Hall effect current sensor;

[0076] Function: Monitor whether the current consumption of the equipment is within the normal range and identify faults in the motor or its components.

[0077] Response speed sensor:

[0078] Type: photoelectric sensor or contact switch;

[0079] Function: Detect the time delay of the device's response to instructions and analyze the device's response performance in different states.

[0080] Pressure sensor:

[0081] Type: strain gauge pressure sensor;

[0082] Function: Monitor pressure changes in hydraulic or pneumatic systems to ensure the response speed and efficiency of the equipment under different loads.

[0083] Temperature sensor:

[0084] Type: Thermocouple or thermistor (RTD);

[0085] Function: Real-time monitoring of the temperature of equipment working parts, timely detection of temperature anomalies, and prevention of equipment damage or failure.

[0086] Non-contact infrared temperature sensor:

[0087] Type: infrared temperature sensor;

[0088] Function: Able to quickly measure the surface temperature of equipment without contact, suitable for high temperature or easily damaged occasions.

[0089] Load Cell:

[0090] Type: Strain gauge load cell;

[0091] Function: Monitor the actual load carried by the equipment to ensure that the equipment operates within the load range and prevent overloading.

[0092] Force sensor:

[0093] Type: piezoelectric force sensor;

[0094] Function: Monitor the force applied to the equipment to determine whether the equipment exceeds the safe load during operation.

[0095] Distance sensor:

[0096] Type: Ultrasonic sensor or laser rangefinder;

[0097] Function: Measure the space distance around the equipment to ensure that the equipment has enough space buffer during operation to avoid malfunctions caused by insufficient space.

[0098] Position Sensor:

[0099] Type: Photoelectric sensor or contact position switch;

[0100] Function: Monitor the equipment position to ensure that the equipment is within the predetermined operating range and prevent equipment damage due to lack of space.

[0101] Furthermore, the data processing module 2 includes:

[0102] Preprocessing submodule, first feature extraction submodule, second feature extraction submodule and fusion submodule;

[0103] The preprocessing submodule is used to perform time series alignment, outlier detection and outlier correction on the multi-source data to obtain preprocessed multi-source data, wherein the preprocessed multi-source data includes: preprocessed operating parameters, preprocessed instantaneous load parameters, preprocessed environmental parameters and preprocessed electrical states; the first feature extraction submodule is used to extract characteristic signals of the preprocessed operating parameters, preprocessed instantaneous load parameters and preprocessed environmental parameters to obtain a parameter feature signal set; the second feature submodule is used to use Clark transform and Park transform to extract voltage signals in the preprocessed electrical state to obtain direct-axis and quadrature-axis voltages; the fusion submodule is used to perform feature fusion on the parameter feature signal set and the direct-axis and quadrature-axis voltages to obtain input data.

[0104] Specifically, data input: receiving multi-source data from the data acquisition module 1, including operating parameters, instantaneous load parameters, environmental parameters and electrical status.

[0105] Time series alignment steps:

[0106] Determine the timestamp for each data source.

[0107] Align the time series of each data source based on the minimum time interval (such as sampling frequency) to ensure that data from different sources can be compared at the same time point.

[0108] Outlier detection: Use statistical methods (such as z-score and IQR methods) to detect outliers; mark values ​​outside the normal range and record those points that may be affected by noise.

[0109] Outlier correction: Replace detected outliers, such as using interpolation methods (such as linear interpolation or polynomial interpolation) to fill missing values; ensure that the preprocessed data is smooth and has no mutations. The output includes preprocessed operating parameters, instantaneous load parameters, environmental parameters, and electrical status.

[0110] The workflow of the first feature extraction submodule is as follows: use signal processing technology (such as Fourier transform, wavelet transform) to extract frequency domain or time domain features; extract features including mean, standard deviation, maximum value, minimum value, peak factor, etc.; summarize these feature signals to form a parameter feature signal set.

[0111] The workflow of the second feature extraction submodule is as follows: apply Clark transform (α-β transform) and Park transform (dq transform) to the voltage signal; convert the three-phase voltage signal into direct-axis (d-axis) and quadrature-axis (q-axis) voltage signals; and output the obtained direct-axis and quadrature-axis voltages as feature signals for subsequent feature fusion.

[0112] The workflow of the fusion submodule is as follows: fuse the parameter feature signal set and the electrical feature signal in a certain way (such as weighted averaging, splicing); determine the fusion method to ensure that the new features can represent multi-dimensional data; and output the final input data for subsequent processing by the health assessment module 3.

[0113] Furthermore, the expression of the parameter characteristic signal set is:

[0114] ;

[0115] Among them, n is the number of monitoring types of equipment operating parameters, m is the number of monitoring types of environmental parameters, is the parameter characteristic signal set, is the time series of the i-th operating parameter, is the mean of the time series of the operating parameters after time alignment, is the standard deviation of the time series of the operating parameters after time alignment, is the weight coefficient, is the nonlinear amplification coefficient, is the instantaneous load value, is the load safety threshold, is the jth environmental parameter, For environmental reference benchmarks, The environmental coupling coefficient is the time series of operating parameters including speed, temperature, and pressure. The weight coefficient is dynamically adjusted through an attention mechanism to reflect parameter importance. The instantaneous load value includes peak current and mechanical torque. The load safety threshold is used to avoid the logarithmic function's sensitivity to zero values. The nonlinear amplification factor is used to highlight over-limit risks. Environmental parameters include humidity, dust concentration, and vibration amplitude. The environmental reference benchmark is used to quantify the degree of anomaly. The environmental coupling coefficient is calibrated using the device's material properties.

[0116] Specifically, through differentiated processing of standardization (the first item), nonlinear amplification (the second item), and environmental coupling (the third item), the dimensions are unified and the physical meaning is retained; dynamic weights achieve adaptive feature selection and avoid manual experience intervention.

[0117] The logarithmic function and product term design have exponential sensitivity to load overload and environmental deterioration, improving early warning capabilities.

[0118] Furthermore, the expression of the input data is:

[0119] ;

[0120] in, is the weight matrix, and are the direct-axis / quadrature-axis voltages extracted by Clark transform and Park transform, respectively. is the electrical characteristic weight matrix, is the bias term, is a non-linear activation function.

[0121] The weight matrix is ​​optimized through adversarial training to suppress redundant features; the bias term is used to compensate for the device baseline error; and the nonlinear activation function is used to filter negative noise interference.

[0122] The implicit rules of mechanical parameters and electrical characteristics are learned separately through independent weight matrices to avoid inter-modal interference; the ReLU function enhances sparsity and focuses on key fault signals.

[0123] The direct-axis voltage reflects the steady-state deviation of the electrical system, and the quadrature-axis voltage characterizes the transient response. They complement the mechanical parameter characteristics and cover all operating conditions of the equipment.

[0124] Furthermore, the health assessment module 3 includes:

[0125] Dimensionality reduction submodule, model building submodule, health index calculation submodule, health assessment submodule and identification submodule;

[0126] The dimensionality reduction submodule is used to reduce the dimensionality of the input data to obtain the input data after dimensionality reduction. The model construction submodule is used to construct a health index calculation model based on the machine learning model. The health index calculation submodule is used to obtain the output of the health index calculation model and obtain a health index based on the output. The health assessment submodule is used to determine the current health level based on the health index, wherein the health level is divided into excellent, good, medium, and poor. The identification submodule is used to determine the type of health problem based on the health index.

[0127] Specifically, the dimensionality reduction method is: select an appropriate dimensionality reduction algorithm, such as principal component analysis (PCA), t-distributed stochastic neighbor embedding (t-SNE), or linear discriminant analysis (LDA); use the selected dimensionality reduction algorithm to process the input data, reduce the feature dimensions, and retain the most important information; output the reduced dimensionality input data for subsequent model construction.

[0128] Health assessment process: Set the criteria for dividing health levels, such as:

[0129] Excellent: health index is between 80%-100%;

[0130] Good: health index is between 60%-79%;

[0131] Medium: health index is between 40%-59%;

[0132] Poor: health index is between 0-39%;

[0133] Determine the health level of the current device based on the input health index.

[0134] Output health level information for reference by operation and maintenance personnel.

[0135] Identification of health problem types:

[0136] Set conditional rules based on health level, for example:

[0137] If the health level is "Excellent", there are no health problems.

[0138] If the health level is "good" or "moderate", possible health problems are identified.

[0139] If the health level is "poor", it is clearly identified as a major failure of the equipment.

[0140] Furthermore, the expression of the health index calculation model is:

[0141] ;

[0142] in, is the input data after dimensionality reduction, For the The encoder output of the autoencoder, is the dynamic attention weight, is the output of the health index calculation model, is a Sigmoid function that maps the output to the interval [0,1][0,1], representing the health index (HI=1 is the best state), K is the total number of autoencoder groups, k is the index of the autoencoder, and j is the index of the environmental parameter.

[0143] in, ;

[0144] is the time window characteristic of the historical health index (such as sliding mean, trend slope), is the trainable weight matrix.

[0145] Furthermore, the identification submodule includes:

[0146] separation units, extraction units, distribution units, and classification units;

[0147] The classification unit is used to receive the input data after dimensionality reduction and split the input data after dimensionality reduction into three independent feature sets. The extraction unit is used to extract key features of the three independent feature sets to obtain key feature sets. The allocation unit is used to perform weight allocation on the key feature sets to obtain allocated feature sets. The classification unit is used to obtain the type of health problem based on the trained classifier and the allocated feature sets.

[0148] Specifically, based on the relevance of device features (such as working status, load conditions, and environmental conditions), the reduced-dimensional input data is split into three independent feature sets. For example:

[0149] The first feature set: features related to the working status of the equipment (such as vibration, temperature, etc.);

[0150] Second feature set: features related to load conditions (such as torque, continuous load, etc.);

[0151] The third feature set: features related to environmental conditions (such as humidity, pressure, etc.);

[0152] Output these three independent feature sets for subsequent analysis.

[0153] Apply feature selection algorithms (such as variance-based selection, L1 regularization method, importance of tree model, etc.) to each independent feature set to extract key features.

[0154] The most representative features are identified and extracted from each feature set to form a key feature set.

[0155] Output key feature set, which contains the most important features that can characterize the health status of the equipment.

[0156] Assign a weight to each key characteristic based on its impact on the health of the equipment. The weight can be determined based on historical data analysis or through expert evaluation.

[0157] The allocation method can use a normalization process to ensure that all feature weights sum to 1, or define a specific weighting strategy (such as normal distribution weighting).

[0158] Output the assigned feature set, which contains weighted key feature data.

[0159] The assigned feature set is input into a trained classifier, which can be a support vector machine, random forest, decision tree, or neural network.

[0160] The classifier infers based on the input feature set and outputs the type of health problem (such as abnormal equipment working status, equipment load exceeding the limit, high temperature, reduced response speed, etc.).

[0161] Output the identified health problem types for reference and further processing by operation and maintenance personnel.

[0162] Furthermore, the operation and maintenance work order generation includes:

[0163] Work order template creation:

[0164] Prepare a standard work order format based on the organization's operation and maintenance work order template, including key information fields that need to be filled in, such as:

[0165] Work order number, equipment identification, fault description, fault location information, prompt measures, and generation time;

[0166] Fill in the work order information:

[0167] Fill the collected fault location information into the work order template to ensure that all important information is recorded;

[0168] Work order storage and distribution:

[0169] The generated operation and maintenance work order is saved in the system database and triggers notifications from relevant departments (such as via email, text message, etc.) so that the operation and maintenance personnel can respond in a timely manner.

[0170] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0171] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. An airport equipment health monitoring system, characterized in that: include: The data acquisition module is configured at the key nodes of the target device to collect multi-source data of the device in real time; A data processing module, configured to integrate the multi-source data to obtain input data; A health assessment module, which uses input data to generate a device health index and health level based on a machine learning model fused with multimodal data, and identifies health issue types, including abnormal device operating status, reduced device response speed, excessive device operating temperature, excessive device load, and insufficient device space margin. The early warning response module is used to trigger a targeted early warning signal based on the equipment health index and generate an operation and maintenance work order containing fault location information.

2. The airport equipment health monitoring system according to claim 1, characterized in that: The multi-source data includes: operating parameters, instantaneous load parameters, environmental parameters and electrical status.

3. The airport equipment health monitoring system according to claim 1, characterized in that: The data acquisition module includes: Node determination submodule, node identification submodule, sensor placement submodule; The node determination submodule is used to determine the key nodes of the target device, the node identification submodule is used to identify the type of the key node and obtain an identification result, and the sensor placement submodule is used to place corresponding sensors at the key node positions according to the identification result.

4. The airport equipment health monitoring system according to claim 2, characterized in that: The data processing module includes: Preprocessing submodule, first feature extraction submodule, second feature extraction submodule and fusion submodule; The preprocessing submodule is used to perform time series alignment, outlier detection and outlier correction on the multi-source data to obtain preprocessed multi-source data, wherein the preprocessed multi-source data includes: preprocessed operating parameters, preprocessed instantaneous load parameters, preprocessed environmental parameters and preprocessed electrical states; the first feature extraction submodule is used to extract characteristic signals of the preprocessed operating parameters, preprocessed instantaneous load parameters and preprocessed environmental parameters to obtain a parameter feature signal set; the second feature submodule is used to use Clark transform and Park transform to extract voltage signals in the preprocessed electrical state to obtain direct-axis and quadrature-axis voltages; the fusion submodule is used to perform feature fusion on the parameter feature signal set and the direct-axis and quadrature-axis voltages to obtain input data.

5. The airport equipment health monitoring system according to claim 4, characterized in that: The expression of the parameter characteristic signal set is: ； in, is the parameter characteristic signal set, is the time series of the i-th operating parameter, is the mean of the time series of the operating parameters after time alignment, is the standard deviation of the time series of the operating parameters after time alignment, is the weight coefficient, is the nonlinear amplification coefficient, is the instantaneous load value, is the load safety threshold, is the jth environmental parameter, For environmental reference benchmarks, is the environmental coupling coefficient, n is the number of monitoring types of equipment operating parameters, and m is the number of monitoring types of environmental parameters.

6. The airport equipment health monitoring system according to claim 4, characterized in that: The expression of the input data is: ； in, is the weight matrix, and are the direct-axis / quadrature-axis voltages extracted by Clark transform and Park transform, respectively. is the electrical characteristic weight matrix, is the bias term, is a non-linear activation function.

7. The airport equipment health monitoring system according to claim 4, characterized in that: The health assessment module includes: Dimensionality reduction submodule, model building submodule, health index calculation submodule, health assessment submodule and identification submodule; The dimensionality reduction submodule is used to reduce the dimensionality of the input data to obtain the input data after dimensionality reduction. The model construction submodule is used to construct a health index calculation model based on the machine learning model. The health index calculation submodule is used to obtain the output of the health index calculation model and obtain a health index based on the output. The health assessment submodule is used to determine the current health level based on the health index, wherein the health level is divided into excellent, good, medium, and poor. The identification submodule is used to determine the type of health problem based on the health index.

8. The airport equipment health monitoring system according to claim 7, characterized in that: The expression of the health index calculation model is: ； in, is the input data after dimensionality reduction, For the The encoder output of the autoencoder, is the dynamic attention weight, is the output of the health index calculation model, is the Sigmoid function, K is the total number of autoencoder groups, and k is the index of the autoencoder.

9. The airport equipment health monitoring system according to claim 7, characterized in that: The identification submodule includes: separation units, extraction units, distribution units, and classification units; The classification unit is used to receive the input data after dimensionality reduction and split the input data after dimensionality reduction into three independent feature sets. The extraction unit is used to extract key features of the three independent feature sets to obtain key feature sets. The allocation unit is used to perform weight allocation on the key feature sets to obtain allocated feature sets. The classification unit is used to obtain the type of health problem based on the trained classifier and the allocated feature sets.

Images