Off-line detection method for three-axis atmosphere data subsystem

Abstract

The invention discloses an off-line detection method for a three-axis atmosphere data subsystem. A main control module, a touch display screen, a power supply conditioning module, a test tool module, a communication control module, a backboard power supply module, an alternating current power supply board, a three-axis atmosphere data subsystem test module and an aviation connector are included. By integrating multiple functional modules, the airborne working environment of the three-axis atmospheric data subsystem can be simulated, system performance testing and fault diagnosis can be carried out, integration of testing and diagnosis is achieved, the maintenance time is greatly shortened, and the maintenance efficiency is improved. Meanwhile, through the steps of environment simulation, signal input and data acquisition, performance test and fault analysis, result display and recording and the like, the external field maintenance efficiency is improved, the reliability and accuracy of a test result are improved, a specific fault point can be quickly positioned, intelligent upgrading of detection equipment is realized, and the detection efficiency is improved. And the maintenance requirement of a modern aviation equipment complex system is met.

Description

Technical Field

[0001] The invention belongs to the technical field of aviation three-axis atmospheric data system detection, and in particular relates to an off-line detection method for a three-axis atmospheric data subsystem. Background Art

[0002] The three-axis atmospheric data subsystem is an important avionics device, which is mainly used to measure the aircraft's pressure altitude, airspeed, atmospheric static temperature, angle of attack and other parameters, and is of great significance for improving the aircraft's flight performance and safety. At the same time, due to the complex working principle of the three-axis atmospheric data subsystem and the large number of cross-linked components on board, it is difficult to carry out maintenance work. At present, there is an urgent need for offline detection equipment for the three-axis atmospheric data subsystem to assist in the completion of maintenance and system troubleshooting, and to improve the efficiency of field maintenance and support.

[0003] Therefore, it is necessary to develop a three-axis atmospheric data subsystem tester to simulate the onboard working environment of the three-axis atmospheric data subsystem and assist maintenance personnel to complete offline detection of the three-axis atmospheric data subsystem, thereby enabling the intelligent upgrade of the detection equipment and meeting the maintenance needs of complex systems of modern aviation equipment. Summary of the invention

[0004] The present invention provides an off-line detection method for a three-axis atmospheric data subsystem, which measures indicators and performance, can simulate the airborne working environment of the three-axis atmospheric data subsystem, perform system performance testing and fault diagnosis, realize the integration of testing and diagnosis, greatly reduce maintenance time, and improve field maintenance efficiency. At the same time, the reliability and accuracy of the test results are improved, and the specific fault point can be quickly located, meeting the maintenance requirements of complex systems of modern aviation equipment.

[0005] The above-mentioned purpose of the present invention can be achieved by the following measures. In the first aspect, the present invention provides an offline detection device for a three-axis atmospheric data subsystem, including: a main control module, which is used to control the operation of the entire detection device, receive input signals and output detection results; a touch display screen, connected to the main control module, used to display the operation interface and detection results of the detection device; a power conditioning module, which is powered by AC220V and airborne DC28V, and outputs multiple DC power supplies to power the equipment and test modules; a QXI bus backplane, which provides communication and signal connections for connecting a variety of test modules; a test tooling module, which is used to assist in the angular position accuracy test of the velocity vector sensor; a communication control module, which is used for signal interaction between the main control module and each test module; a backplane power supply module, which provides a variety of DC power sources to be introduced into the bus backplane; an AC power supply board, which is used to provide a variety of AC power signals; a three-axis atmospheric data subsystem test module, which is used to simulate the onboard working environment of the three-axis atmospheric data subsystem; and an aviation connector, which is used to connect to external accessories.

[0006] Furthermore, the test module includes: a temperature field pressure simulator for simulating temperature and static pressure environment; a vector sensor simulator for simulating the solution signal of flight attitude; an atmospheric aircraft simulator for simulating the three-axis atmospheric data computer to receive the azimuth from the velocity vector sensor, display the flight attitude of the aircraft, and feed back to the main control module to judge and display the value; a bus simulation module for simulating and verifying the reception and transmission of ARINC429 bus signals; a data acquisition module for collecting analog signals, AC signals and resistance output by the atmospheric data system; a relay matrix board for testing the output of resources and the connection and disconnection control of measurement signals and the device under test.

[0007] Furthermore, the external accessories include an atmospheric source module for providing an atmospheric pressure altitude, airspeed and ascent and descent rate test environment.

[0008] On the other hand, the present invention also provides an offline detection method for a three-axis atmospheric data subsystem, which is applicable to the above-mentioned device and is characterized in that it comprises the following steps:

[0009] (1) Environmental simulation

[0010] Start the atmospheric source module and the temperature field pressure simulation module, set the pressure altitude, airspeed, ascent and descent rate and temperature parameters, and simulate the airborne working environment;

[0011] Use the vector sensor simulation module to generate solution signals and simulate the aircraft attitude information;

[0012] The simulated aircraft attitude information includes the position information in the three-dimensional coordinate system and the onboard working environment information. Posture =E xyz (P x +P y +P z ), where E xyz Indicates the onboard working environment information, including pressure altitude, airspeed, ascent and descent rate and temperature parameters, P i Represents the position information in a three-dimensional coordinate system;

[0013] (2) Signal input and data acquisition

[0014] Send signals that comply with the ARINC429 protocol to the device under test through the bus simulation module;

[0015] Start the data acquisition module to collect the analog signal, AC signal and resistance signal output by the device under test, and transmit them to the main control module for processing in real time;

[0016] (3) Performance testing and fault analysis

[0017] Analyze the collected data, including comparing it with standard template data;

[0018] Through the fault diagnosis algorithm of the main control module, the performance deviation or fault source of the tested device is identified and the fault point is marked;

[0019] (4) Result display and recording

[0020] The test results and fault analysis conclusions are displayed in real time on the touch screen;

[0021] Automatically generate test reports including the model of the tested device, test items, actual collected values, diagnosis results and maintenance suggestions.

[0022] Furthermore, the fault diagnosis algorithm includes a data preprocessing module, which normalizes the collected signals before fault diagnosis to ensure data quality and consistency; an anomaly detection module, which is used to identify abnormal patterns of signals; a fault classification module, which is used to further determine the specific fault type; a fault risk assessment module and a fault trend prediction module.

[0023] Furthermore, the data preprocessing module includes the following steps:

[0024] Signal filtering: Use a sliding average filter to remove high-frequency noise; Use wavelet transform denoising method to process temperature and pressure signals and retain low-frequency components; Signal alignment: Synchronize timestamps of multi-source signals to ensure time consistency of different signal acquisitions; Standardization: Standardize all signals to the [0,1] interval, the formula is:

[0025]

[0026] Furthermore, the anomaly detection module comprises the following steps:

[0027] Input data: Input multidimensional feature signal X = [x1, x2, x3…, x n ,P Posture ], such as air pressure signals, temperature signals, velocity vector signals, and simulated aircraft attitude information;

[0028] Module training: The autoencoder is used to capture the characteristic distribution of normal signals and determine whether the signal is abnormal through reconstruction error; the encoder part compresses the input signal into a low-dimensional latent space and learns the characteristic distribution of normal signals; the decoder part reconstructs the latent space features back to the original signal dimension;

[0029] Use autoencoders to perform unsupervised training on normal signals to minimize reconstruction errors

[0030]

[0031] where x i is the input signal, is the reconstruction signal;

[0032] Abnormality determination: The error after the model reconstructs the input signal:

[0033]

[0034] When the reconstruction error exceeds the preset threshold, it is marked as an abnormal signal.

[0035] Furthermore, the specific steps of the fault classification module are:

[0036] Input: the degree of signal deviation, that is, the difference between the measured value and the reference value; the amplitude of signal fluctuation, that is, the standard deviation and the difference between the maximum and minimum values;

[0037] Module algorithm: Use the random forest algorithm model to define multiple decision trees, each of which is trained based on a different feature subset; based on the classification results of all decision trees, the majority vote is taken as the final fault classification;

[0038] Output: The specific type of failure.

[0039] Furthermore, the fault risk assessment module uses a Bayesian network to conduct a comprehensive assessment of the fault risk.

[0040] The probability of each fault occurring is:

[0041]

[0042] The output is a risk score for each failure, in the range [0,1].

[0043] Furthermore, the fault trend prediction module uses a time series model to analyze the signal change trend in multiple detections.

[0044] Beneficial Effects

[0045] The above technical solution of the present invention has the following advantages compared with the prior art:

[0046] 1. The present invention integrates multiple functional modules to simulate the airborne working environment of the three-axis atmospheric data subsystem, maps the actual geographical location and environment into a three-dimensional coordinate system, and performs system performance testing and fault diagnosis by simulating the environment and equipment operation in the actual operation process, thereby realizing the integration of testing and diagnosis, greatly reducing maintenance time, and improving field maintenance efficiency.

[0047] 2. The present invention can accurately collect analog signals, AC signals and resistance signals through the data acquisition module, and perform high-precision anomaly detection and fault classification in combination with the artificial intelligence model, thereby improving the reliability and accuracy of the test results and quickly locating the specific fault point.

[0048] 3. The present invention adopts artificial intelligence algorithm to analyze the dynamic change trend of the signal, which can identify potential fault risks in advance, provide data support for early intervention of faults, avoid safety hazards caused by further expansion of faults, realize the intelligent upgrade of detection equipment, and meet the maintenance needs of complex systems of modern aviation equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is the system layout diagram of the offline testing equipment for this application;

[0050] Figure 2 The hardware module layout diagram of the QXI chassis for this application;

[0051] Figure 3 This is a flow chart of the offline detection method of the offline detection device of this application;

[0052] Figure 4 This is the detection flow chart of the anomaly detection module of this application. DETAILED DESCRIPTION

[0053] The present invention will be further described below in conjunction with the embodiments.

[0054] Example 1: Figure 1-2 As shown, the present invention provides an offline detection device for a three-axis atmospheric data subsystem, which mainly includes: a main control module, which is used to control the operation of the entire detection device, receive input signals and output detection results. It is the core component of the tester, used for the input and output of signals of the entire device, and realizes the coordination and control of various functions. At the same time, it communicates with the communication control module through the RS485 bus, sends control commands, and returns the test data of each functional module for displaying the test results;

[0055] A touch screen connected to the main control module, used to display the operation interface and test results of the test equipment;

[0056] The power conditioning module is powered by AC220V and onboard DC28V, and outputs multiple DC power supplies to power the equipment and test modules; the tester adopts AC 220V 50HZ or DC 28V dual power supply optional power supply mode, and converts the input AC 220V50HZ into DC voltage through the AC to DC module. This voltage is selected in parallel with the DC 28V input power supply, and the 28V DC input is prioritized. The selected voltage becomes 28V DC and 12V DC 2-way output to power the entire test system and the device under test;

[0057] QXI bus backplane provides communication and signal connections for connecting various test modules; the bus backplane is mainly used for signal transfer, transferring the control signal of the conditioned output of the communication control module to the communication control module, backplane power module, data acquisition board, 429 bus simulator, relay matrix board, temperature field pressure simulator, vector sensor simulator, atmospheric machine simulator, 115V AC power board, 36V AC power board and other hardware test resource modules through 12 rows of standard European 48-pin sockets, and each row of sockets is point-to-point connected; it also provides 28V, 15V, -15V and 5V power supplies, three serial communication buses, 16 data lines, and 4 address lines;

[0058] The test fixture module is used to assist in the angular position accuracy test of the velocity vector sensor; it provides the velocity vector sensor test angular position, which is used to test the sinφ1, cosφ1, sinφ1+φ2, cosφ1+φ2 output accuracy of the velocity vector sensor at different angles;

[0059] The communication control module is used for signal interaction between the main control module and each test module. It converts the serial port commands issued by the main control module into RS485 bus control commands defined by the bus backplane, and sends them to other functional modules to complete signal control, switching and test measurement functions. It also converts the test data fed back from the backplane into serial port commands and feeds them back to the main control module for measurement display.

[0060] The backplane power module provides multiple DC power supplies for the bus backplane. It converts the input 12V DC power into 5V, 15V, and -15V DC power supplies for the bus backplane for use by other functional modules. The 28V DC power supply is filtered and then introduced into the bus backplane for use by other functional modules. It controls the output of 28V, 15V, -15V, and 5V DC power supplies for component testing. It can read the current consumption of the four output power supplies.

[0061] AC power board, used to provide a variety of AC power signals;

[0062] A three-axis atmospheric data subsystem test module is used to simulate the onboard working environment of the three-axis atmospheric data subsystem; and an aviation connector is used to connect with external accessories.

[0063] Furthermore, the test module includes: a temperature field pressure simulator for simulating temperature and static pressure environment; a vector sensor simulator for simulating the solution signal of flight attitude; an atmospheric aircraft simulator for simulating the three-axis atmospheric data computer to receive the azimuth from the velocity vector sensor, display the flight attitude of the aircraft, and feed back to the main control module to judge and display the value; a bus simulation module for simulating and verifying the reception and transmission of ARINC429 bus signals; a data acquisition module for collecting analog signals, AC signals and resistance output by the atmospheric data system; a relay matrix board for testing the output of resources and the connection and disconnection control of measurement signals and the device under test.

[0064] Furthermore, the external accessories include an atmospheric source module for providing an atmospheric pressure altitude, airspeed and ascent and descent rate test environment.

[0065] Example 2: Figure 3 As shown, the present invention also provides an offline detection method for a three-axis atmospheric data subsystem, which is applicable to the device described in claim 1 and is characterized in that it comprises the following steps:

[0066] (1) Environmental simulation

[0067] Start the atmospheric source module and the temperature field pressure simulation module, set the pressure altitude, airspeed, ascent and descent rate and temperature parameters, and simulate the airborne working environment;

[0068] Use the vector sensor simulation module to generate solution signals and simulate the aircraft attitude information;

[0069] The simulated aircraft attitude information includes the position information in the three-dimensional coordinate system and the onboard working environment information. Posture =E xyz (P x +P y +P z ), where E xyz Indicates the onboard working environment information, including pressure altitude, airspeed, ascent and descent rate and temperature parameters, P i Represents the position information in a three-dimensional coordinate system;

[0070] (2) Signal input and data acquisition

[0071] Send signals that comply with the ARINC429 protocol to the device under test through the bus simulation module;

[0072] Start the data acquisition module to collect the analog signal, AC signal and resistance signal output by the device under test, and transmit them to the main control module for processing in real time;

[0073] (3) Performance testing and fault analysis

[0074] Analyze the collected data, including comparing it with standard template data;

[0075] Through the fault diagnosis algorithm of the main control module, the performance deviation or fault source of the tested device is identified and the fault point is marked;

[0076] (4) Result display and recording

[0077] The test results and fault analysis conclusions are displayed in real time on the touch screen;

[0078] Automatically generate test reports including the model of the tested device, test items, actual collected values, diagnosis results and maintenance suggestions.

[0079] Furthermore, the fault diagnosis algorithm includes a data preprocessing module, which normalizes the collected signals before fault diagnosis to ensure data quality and consistency; an anomaly detection module, which is used to identify abnormal patterns of signals; a fault classification module, which is used to further determine the specific fault type; a fault risk assessment module and a fault trend prediction module.

[0080] Furthermore, the data preprocessing module includes the following steps:

[0081] Signal filtering: Use a sliding average filter to remove high-frequency noise; Use wavelet transform denoising method to process temperature and pressure signals and retain low-frequency components; Signal alignment: Synchronize timestamps of multi-source signals to ensure time consistency of different signal acquisitions; Standardization: Standardize all signals to the [0,1] interval, the formula is:

[0082]

[0083] Furthermore, if Figure 4 As shown, the anomaly detection module includes the following steps:

[0084] Input data: Input multidimensional feature signal X = [x1, x2, x3…, x n ,P Posture ], such as air pressure signals, temperature signals, velocity vector signals, and simulated aircraft attitude information;

[0085] Module training: The autoencoder is used to capture the characteristic distribution of normal signals and determine whether the signal is abnormal through reconstruction error; the encoder part compresses the input signal into a low-dimensional latent space and learns the characteristic distribution of normal signals; the decoder part reconstructs the latent space features back to the original signal dimension;

[0086] Use autoencoders to perform unsupervised training on normal signals to minimize reconstruction errors

[0087]

[0088] where x i is the input signal, is the reconstruction signal;

[0089] Abnormality determination: The error after the model reconstructs the input signal:

[0090]

[0091] When the reconstruction error exceeds the preset threshold, it is marked as an abnormal signal.

[0092] Furthermore, the specific steps of the fault classification module are:

[0093] Input: the degree of signal deviation, that is, the difference between the measured value and the reference value; the amplitude of signal fluctuation, that is, the standard deviation and the difference between the maximum and minimum values;

[0094] Module algorithm: Use the random forest algorithm model to define multiple decision trees, each of which is trained based on a different feature subset; based on the classification results of all decision trees, the majority vote is taken as the final fault classification;

[0095] Output: The specific type of failure.

[0096] Furthermore, the fault risk assessment module uses a Bayesian network to conduct a comprehensive assessment of the fault risk.

[0097] The probability of each fault occurring is:

[0098]

[0099] The output is a risk score for each failure, in the range [0,1].

[0100] Furthermore, the fault trend prediction module uses a time series model to analyze the signal change trend in multiple detections.

[0101] In Example 3, in the detection step of the anomaly detection module, the anomaly threshold can be determined by the reconstruction error distribution of the training set: when the autoencoder is used for anomaly detection, we need to determine a reasonable threshold through the training process of normal data to distinguish normal signals from abnormal signals.

[0102] 1. Selection of training set

[0103] Definition of training set: The training data set should include signal acquisition data of the system under normal working conditions, without any known faults or anomalies.

[0104] Data Type:

[0105] Static signals (such as steady-state values ​​of temperature, air pressure, etc.)

[0106] Dynamic signals (such as the change curve of velocity vector signal)

[0107] 2. Model training

[0108] Input: Input the training data into the autoencoder, and the input feature is multi-dimensional signal data, for example: X = [x1, x2, x3…, x n ], xi is the data point of a single signal dimension

[0109] Encoding-Decoding: Through the autoencoder model, a low-dimensional representation of normal data is learned and the original data is reconstructed.

[0110] Reconstruction error calculation:

[0111] For each input data point x i , calculate its reconstruction value with the model output The error between:

[0112]

[0113] 3. Statistical Analysis of Reconstruction Error Distribution

[0114] Statistics of the error set:

[0115] Calculate the reconstruction error of all data points in the entire training set to form an error set ε = {∈1,∈2,…,∈ m},in:

[0116] Error distribution analysis:

[0117] Plot the error distribution histogram. Usually the reconstruction error of the training data will show a normal distribution or a form close to a normal distribution.

[0118] Calculate the mean μ and standard deviation σ of the error distribution:

[0119]

[0120] Determine the anomaly threshold:

[0121] The threshold is set as the upper bound of the error distribution, which is usually defined as: Threshold = μ + k·σ, where k is a constant, usually k = 3 (i.e., 3 times the standard deviation, covering about 99.7% of the data points in the normal distribution).

[0122] The threshold represents the error range of normal signals, and data points exceeding this range are marked as abnormal.

[0123] 4. Abnormal determination in practical applications

[0124] Real-time error calculation:

[0125] In practical applications, for each new input signal point x new , calculate its reconstruction error:

[0126] Compare with threshold:

[0127] If ∈ new >Threshold, then mark x new It is an abnormal signal.

[0128] Otherwise, the signal is considered normal.

[0129] 5. Dynamic threshold adjustment

[0130] Retrain based on new normal data:

[0131] As the system operates normally, new normal signal data can be collected to fine-tune the model and update the error distribution and threshold.

[0132] Environmental adaptability adjustment:

[0133] If the working environment changes significantly (such as the temperature or pressure range increases), the threshold kkk may need to be readjusted to adapt to the new conditions.

[0134] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. An offline detection method for a three-axis atmospheric data subsystem, comprising an offline detection device for the three-axis atmospheric data subsystem, wherein the offline detection device comprises a main control module for controlling the operation of the entire detection device, receiving input signals and outputting detection results; a touch screen connected to the main control module for displaying an operation interface of the detection device and detection results; The power conditioning module is powered by AC220V and onboard DC28V, and outputs multiple DC power supplies to power the equipment and test modules; the QXI bus backplane provides communication and signal connections for connecting various test modules; Test fixture module, used to assist in the angular position accuracy test of the velocity vector sensor; Communication control module, used for signal interaction between the main control module and each test module; backplane power supply module, providing a variety of DC power supply to the bus backplane; AC power board, used to provide a variety of AC power signals; three-axis atmospheric data subsystem test module, used to simulate the onboard working environment of the three-axis atmospheric data subsystem; and aviation connectors for connecting to external accessories; The test fixture module includes: a temperature field pressure simulator for simulating temperature and static pressure environment; a vector sensor simulator for simulating the solution signal of flight attitude; an atmospheric machine simulator for simulating the three-axis atmospheric data computer to receive the azimuth from the velocity vector sensor, display the flight attitude of the aircraft, and feed back to the main control module to judge and display the value; a bus simulation module for simulating and verifying the reception and transmission of ARINC429 bus signals; a data acquisition module for collecting analog signals, AC signals and resistance output by the atmospheric data system; a relay matrix board for testing the output of resources and the connection and disconnection control of the measurement signal and the tested device; the external accessories include an atmospheric source module for providing a pressure altitude, airspeed and lift rate test environment; it is characterized in that it includes the following steps: (1) Environmental simulation Start the atmospheric source module and the temperature field pressure simulation module, set the pressure altitude, airspeed, ascent and descent rate and temperature parameters, and simulate the airborne working environment; Use the vector sensor simulation module to generate solution signals and simulate the aircraft attitude information; The simulated aircraft attitude information includes the position information in the three-dimensional coordinate system and the onboard working environment information. Posture =E xyz (P x +P y +P z ), where E xyz Indicates the onboard working environment information, including pressure altitude, airspeed, ascent and descent rate and temperature parameters, P i Represents the position information in a three-dimensional coordinate system; (2) Signal input and data acquisition Send signals that comply with the ARINC429 protocol to the device under test through the bus simulation module; Start the data acquisition module to collect the analog signal, AC signal and resistance signal output by the device under test, and transmit them to the main control module for processing in real time; (3) Performance testing and fault analysis Analyze the collected data, including comparing it with standard template data; Through the fault diagnosis algorithm of the main control module, the performance deviation or fault source of the tested device is identified and the fault point is marked; (4) Result display and recording The test results and fault analysis conclusions are displayed in real time on the touch screen; Automatically generate test reports, including the model of the tested part, test items, actual collected values, diagnostic results and maintenance suggestions; The fault diagnosis algorithm includes a data preprocessing module, which normalizes the collected signals before fault diagnosis to ensure data quality and consistency; an anomaly detection module, which is used to identify abnormal patterns of signals; a fault classification module, which is used to further determine the specific fault type; a fault risk assessment module and a fault trend prediction module; The data preprocessing module comprises the following steps: Signal filtering: Use a sliding average filter to remove high-frequency noise; Use wavelet transform denoising method to process temperature and pressure signals and retain low-frequency components; Signal alignment: Synchronize the timestamps of multi-source signals to ensure the time consistency of different signal acquisitions; Standardization: Standardize all signals to the [0,1] interval, the formula is:

2. The method according to claim 1, characterized in that The anomaly detection module comprises the following steps: Input data: Input multidimensional feature signal X = [x1, x2, x3…, x n ,P Posture ], such as air pressure signals, temperature signals, velocity vector signals, and simulated aircraft attitude information; Module training: The autoencoder is used to capture the characteristic distribution of normal signals and determine whether the signal is abnormal through reconstruction error; the encoder part compresses the input signal into a low-dimensional latent space and learns the characteristic distribution of normal signals; the decoder part reconstructs the latent space features back to the original signal dimension; Use autoencoders to perform unsupervised training on normal signals to minimize reconstruction errors where x i is the input signal, is the reconstruction signal; Abnormality determination: The error after the model reconstructs the input signal: When the reconstruction error exceeds the preset threshold, it is marked as an abnormal signal.

3. The method according to claim 2, characterized in that The specific steps of the fault classification module are: Input: the degree of signal deviation, that is, the difference between the measured value and the reference value; the amplitude of signal fluctuation, that is, the standard deviation and the difference between the maximum and minimum values; Module algorithm: Use the random forest algorithm model to define multiple decision trees, each of which is trained based on a different feature subset; based on the classification results of all decision trees, the majority vote is taken as the final fault classification; Output: The specific type of failure.

4. The method according to claim 3, characterized in that The fault risk assessment module uses a Bayesian network to conduct a comprehensive assessment of the fault risk. The probability of each fault occurring is: The output is a risk score for each failure, in the range [0,1].

5. The method according to claim 4, characterized in that The fault trend prediction module uses a time series model to analyze the signal change trend in multiple detections.

Images