CN117727449A - Evaluation method for healthy flight time of civil aviation pilot - Google Patents
Evaluation method for healthy flight time of civil aviation pilot Download PDFInfo
- Publication number
- CN117727449A CN117727449A CN202410173526.3A CN202410173526A CN117727449A CN 117727449 A CN117727449 A CN 117727449A CN 202410173526 A CN202410173526 A CN 202410173526A CN 117727449 A CN117727449 A CN 117727449A
- Authority
- CN
- China
- Prior art keywords
- pilot
- flight
- physical state
- health
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 25
- 230000036541 health Effects 0.000 claims abstract description 162
- 238000013210 evaluation model Methods 0.000 claims abstract description 15
- 238000013135 deep learning Methods 0.000 claims abstract description 8
- 238000000034 method Methods 0.000 claims description 63
- 238000012549 training Methods 0.000 claims description 31
- 230000004617 sleep duration Effects 0.000 claims description 13
- 238000005259 measurement Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 8
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 4
- 239000008280 blood Substances 0.000 claims description 4
- 210000004369 blood Anatomy 0.000 claims description 4
- 230000036760 body temperature Effects 0.000 claims description 4
- 229910052760 oxygen Inorganic materials 0.000 claims description 4
- 239000001301 oxygen Substances 0.000 claims description 4
- 230000007958 sleep Effects 0.000 claims description 4
- 230000004399 eye closure Effects 0.000 claims description 3
- 230000036314 physical performance Effects 0.000 claims description 3
- 238000013526 transfer learning Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 5
- 230000002411 adverse Effects 0.000 abstract description 4
- 238000012545 processing Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 16
- 230000008859 change Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 230000010006 flight Effects 0.000 description 3
- 230000014509 gene expression Effects 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000001815 facial effect Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 210000000744 eyelid Anatomy 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000008821 health effect Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000000819 phase cycle Methods 0.000 description 1
- 230000004622 sleep time Effects 0.000 description 1
- 230000036578 sleeping time Effects 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an evaluation method of healthy flight time of a civil aviation pilot, which relates to the technical field of data processing, and comprises the steps of constructing a pilot physical state prediction database DT and a health value evaluation database DR, and learning data in the pilot physical state prediction database DT by adopting a deep learning algorithm to obtain a pilot flight phase physical state prediction model; learning data in the health value evaluation database DR by adopting a deep learning algorithm to obtain a pilot health value evaluation model; the physical state data can be predicted through the physical state prediction model of the pilot in the flight stage, the predicted physical state data is evaluated through the health numerical evaluation model of the pilot, and finally, the optimal flight time which can be qualified on the premise of not affecting the health can be determined, so that the adverse effect of the improper flight time on the health of the pilot is prevented.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method for evaluating the healthy flight time of a civil aviation pilot.
Background
In the field of modern civil aviation, airlines offer global services that enable people to traverse different continents in a short time, long-distance flights having become an unavoidable phenomenon. Although long-distance flights facilitate global communication, they raise concerns about health effects on pilots. Due to the nature of long-distance flights, including crossing multiple time zones, long duration of flight, and time difference issues that may occur, improper flight duration and mission arrangement may significantly affect the physical health of the pilot
However, in the field of civil aviation today, countermeasures regarding the duration of flight that may negatively affect the health of the pilot are limited to intervention in health interventions only when the pilot has obvious symptoms after completion of the flight mission. This intervention is in fact posterior and the health of the pilot has been affected to some extent.
Disclosure of Invention
The invention aims to provide an evaluation method for the healthy flight time of a civil aviation pilot, which is used for judging the flight mission of which duration the pilot can be qualified in advance by observing the relation between the flight duration of the pilot in the future and the healthy numerical evolution of the pilot. By the assessment method, flight tasks with proper flight time can be arranged more effectively, improper flight time arrangement can be prevented in the form of prior intervention, and the health of pilots is prevented from being adversely affected.
The invention is realized by the following technical scheme:
a method of assessing healthy flight time of a pilot in civil aviation, comprising:
acquiring physical state data and health state data of a civil aviation pilot in continuous time, constructing a pilot physical state prediction database DT according to the physical state data, and constructing a health value evaluation database DR according to the physical state data and the health state data;
taking the pilot physical state prediction database DT as a data support to train a pilot flight phase physical state prediction model; training a pilot health value evaluation model by taking the health value evaluation database DR as data support;
acquiring to-be-evaluated physical state data of a pilot to be evaluated in a ground stage in continuous time, and predicting through a pilot flight stage physical state prediction model to obtain predicted physical state data;
and identifying the predicted physical state data through a pilot health value evaluation model to obtain a predicted health value of a future period, and evaluating the health flight duration according to the predicted health value of the future time.
In one possible embodiment, the physical state data includes pilot basic state data, fatigue state data, and physiological index data;
Wherein the pilot basic state data includes height, weight, age and physical performance indicators;
the fatigue state data comprise that a pilot is in a general fatigue state, a fatigue state which cannot be relieved through rest or a flight is in a normal state;
the physiological index data includes heart rate, blood oxygen saturation, body temperature, and/or sleep duration.
In one possible implementation manner, the method for acquiring fatigue state data includes:
and acquiring a plurality of frames of pilot face images, positioning pilot eye positions through a pilot face key point recognition model trained by transfer learning, and determining the fatigue state type of the pilot by combining a PERCLOS measurement principle and latest sleep duration data.
In one possible embodiment, determining the type of fatigue state of the pilot in combination with PERCLOS measurement principles and recent sleep duration data includes:
based on the eye position of the pilot, determining eye opening and closing information corresponding to each frame of pilot face image; the eye opening and closing information comprises eye opening or eye closing;
according to eye opening and closing information corresponding to all pilot face images, the fatigue index is determined by combining a PERCLOS measurement principle to be:
;
Wherein,PERCLOSthe fatigue index is indicated as a function of the fatigue index,ECthe number of frames representing the eye closure,TFrepresenting the total number of frames of the acquired pilot face image;
judging whether the fatigue index is larger than a preset fatigue threshold value, if so, determining that the fatigue state type is that the pilot is in a fatigue state, otherwise, determining that the fatigue state type is that the pilot is in a normal state;
judging whether the sleep duration data is larger than a preset sleep threshold value, if so, determining that the fatigue state type is a fatigue state which can not be relieved by rest, otherwise, determining that the fatigue state type is a general fatigue state.
In a possible embodiment, with the pilot physical state prediction database DT as a data support, training a pilot flight phase physical state prediction model, comprising:
dividing physical state data of a civil aviation pilot in continuous time into ground-stage physical state data with a fixed length of n and flight-stage physical state data with a fixed length of m by taking the pilot physical state prediction database DT as a data support;
the method comprises the steps of constructing a first initial prediction model through a deep learning algorithm, taking ground stage body state data with a fixed length of n as actual input data of the first initial prediction model, taking flight stage body state data with a fixed length of m as expected prediction data of the first initial prediction model, and training the first initial prediction model to obtain a pilot flight stage body state prediction model.
In one possible embodiment, with the health value assessment database DR as data support, training a pilot health value assessment model, comprising:
taking the health value evaluation database DR as a data support, and extracting the physical state data of the flight stage with the fixed length of m and the physical state data of the flight stage with the fixed length of m, which correspond to the physical state data of the flight stage;
and constructing a second initial prediction model through a deep learning algorithm, taking the physical state data of the flight stage with the fixed length of m as actual input data of the second initial prediction model, taking the physical state data of the flight stage with the fixed length of m corresponding to the physical state data of the flight stage as expected prediction data of the second initial prediction model, and training the second initial prediction model to obtain a pilot health numerical value assessment model.
In one possible embodiment, obtaining physical state data to be evaluated of a pilot to be evaluated in a ground stage on a continuous time, and predicting by a pilot flight stage physical state prediction model to obtain predicted physical state data, including:
acquiring to-be-evaluated physical state data of a pilot to be evaluated in a ground stage on continuous time, and determining to-be-evaluated physical state data on the first n time steps based on the current time on the basis of the obtained to-be-evaluated physical state data on continuous time;
And taking the determined physical state data to be evaluated in the first n time steps as input of a pilot flight stage physical state prediction model to obtain predicted physical state data in the next m time steps.
In one possible embodiment, the method includes identifying predicted physical state data by a pilot health value assessment model, obtaining a predicted health value for a future time period, and assessing a health flight duration based on the predicted health value for the future time period, including:
taking the predicted physical state data of m time steps in the future as the input of a pilot health value evaluation model to obtain predicted health values of m time steps in the future;
based on the predicted health values of the m time steps in the future, performing curve fitting by adopting a spline interpolation method to obtain a time step health value curve;
judging whether a point smaller than a preset health threshold exists in the time step health numerical curve, if so, taking the time corresponding to the first point smaller than the preset health threshold as a first target time, otherwise, determining that the health flight duration is the duration formed by m time steps in the future;
and determining the healthy flight duration to be the duration consisting of k-1 time steps in the future based on the time step k of the first target time.
In one possible embodiment, the method further comprises:
judging whether a point with a slope larger than a preset slope threshold exists in the time step health numerical curve, if so, taking the time corresponding to the point with the first slope larger than the preset slope threshold as a first target time, otherwise, determining that the health flight duration is the duration formed by m time steps in the future;
and determining the healthy flight duration to be the duration consisting of k-1 time steps in the future based on the time step k of the first target time.
In one possible embodiment, the method further comprises:
and when no point smaller than the preset health threshold value and no point with the slope larger than the preset slope threshold value exist in the time step health value curve, continuously acquiring the predicted health value of the future period, and evaluating the health flight duration according to the predicted health value of the future time.
According to the assessment method for the healthy flight time of the civil aviation pilot, the physical state prediction model and the healthy numerical value assessment model of the pilot in the flight stage are obtained by constructing the database and learning the data in the database, and the prediction of the healthy state can be realized through the two models, so that the optimal flight time which can be qualified on the premise of not affecting the health can be determined according to the healthy state, and the adverse effect of the improper flight time on the health of the pilot is prevented.
Drawings
In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings that are needed in the examples will be briefly described below, it being understood that the following drawings only illustrate some examples of the present invention and therefore should not be considered as limiting the scope, and that other related drawings may be obtained from these drawings without inventive effort for a person skilled in the art. In the drawings:
fig. 1 is a flowchart of a method for evaluating healthy flight time of a civil aviation pilot according to an embodiment of the present invention.
Fig. 2 is a schematic structural diagram of an evaluation system for healthy flight time of a civil aviation pilot according to an embodiment of the present invention.
In the drawings, the reference numerals and corresponding part names:
the system comprises a 1-pilot physical state data acquisition module, a 2-pilot flight phase physical state prediction module, a 3-flight phase health value evolution prediction module and a 4-health flight duration evaluation module.
Detailed Description
For the purpose of making apparent the objects, technical solutions and advantages of the present invention, the present invention will be further described in detail with reference to the following examples and the accompanying drawings, wherein the exemplary embodiments of the present invention and the descriptions thereof are for illustrating the present invention only and are not to be construed as limiting the present invention.
As shown in fig. 1, an embodiment of the present invention provides a method for evaluating healthy flight time of a civil aviation pilot, including:
s1, acquiring physical state data and health state data of a civil aviation pilot in continuous time, constructing a pilot physical state prediction database DT according to the physical state data, and constructing a health value evaluation database DR according to the physical state data and the health state data;
s2, training a pilot flight phase physical state prediction model by taking the pilot physical state prediction database DT as a data support; training a pilot health value evaluation model by taking the health value evaluation database DR as data support;
s3, acquiring to-be-evaluated physical state data of a pilot to be evaluated in a ground stage in continuous time, and predicting through a pilot flight stage physical state prediction model to obtain predicted physical state data;
and S4, identifying the predicted physical state data through a pilot health numerical value evaluation model to obtain a predicted health numerical value of a future period, and evaluating the health flight time according to the predicted health numerical value of the future time.
According to the assessment method for the healthy flight time of the civil aviation pilot, the physical state prediction model and the healthy numerical value assessment model of the pilot in the flight stage are obtained by constructing the database and learning the data in the database, and the prediction of the healthy state can be realized through the two models, so that the optimal flight time which can be qualified on the premise of not affecting the health can be determined according to the healthy state, and the adverse effect of the improper flight time on the health of the pilot is prevented.
In one possible embodiment, the physical state data includes pilot basic state data, fatigue state data, and physiological index data;
wherein the pilot basic state data includes height, weight, age and physical performance indicators;
the fatigue state data comprises a fatigue state that the pilot is in a non-fatigue state, a normal fatigue state or cannot be relieved by rest;
the physiological index data includes heart rate, blood oxygen saturation, body temperature, and/or sleep duration.
In one possible implementation manner, the method for acquiring fatigue state data includes:
and acquiring a plurality of frames of pilot face images, positioning pilot eye positions through a pilot face key point recognition model trained by transfer learning, and determining the fatigue state type of the pilot by combining a PERCLOS measurement principle and latest sleep duration data.
In one possible embodiment, determining the type of fatigue state of the pilot in combination with PERCLOS measurement principles and recent sleep duration data includes:
based on the eye position of the pilot, determining eye opening and closing information corresponding to each frame of pilot face image; the eye opening and closing information comprises eye opening or eye closing;
According to eye opening and closing information corresponding to all pilot face images, the fatigue index is determined by combining a PERCLOS measurement principle to be:
；
wherein,PERCLOSthe fatigue index is indicated as a function of the fatigue index,ECthe number of frames representing the eye closure,TFrepresenting the total number of frames of the acquired pilot face image;
judging whether the fatigue index is larger than a preset fatigue threshold value, if so, determining that the fatigue state type is that the pilot is in a fatigue state, otherwise, determining that the fatigue state type is that the pilot is in a normal state;
judging whether the sleep duration data is larger than a preset sleep threshold value, if so, determining that the fatigue state type is a fatigue state which can not be relieved by rest, otherwise, determining that the fatigue state type is a general fatigue state.
In a possible embodiment, with the pilot physical state prediction database DT as a data support, training a pilot flight phase physical state prediction model, comprising:
dividing physical state data of a civil aviation pilot in continuous time into ground-stage physical state data with a fixed length of n and flight-stage physical state data with a fixed length of m by taking the pilot physical state prediction database DT as a data support;
The method comprises the steps of constructing a first initial prediction model through a deep learning algorithm, taking ground stage body state data with a fixed length of n as actual input data of the first initial prediction model, taking flight stage body state data with a fixed length of m as expected prediction data of the first initial prediction model, and training the first initial prediction model to obtain a pilot flight stage body state prediction model.
In one possible embodiment, with the health value assessment database DR as data support, training a pilot health value assessment model, comprising:
taking the health value evaluation database DR as a data support, and extracting the physical state data of the flight stage with the fixed length of m and the physical state data of the flight stage with the fixed length of m, which correspond to the physical state data of the flight stage;
and constructing a second initial prediction model through a deep learning algorithm, taking the physical state data of the flight stage with the fixed length of m as actual input data of the second initial prediction model, taking the physical state data of the flight stage with the fixed length of m corresponding to the physical state data of the flight stage as expected prediction data of the second initial prediction model, and training the second initial prediction model to obtain a pilot health numerical value assessment model.
In one possible embodiment, obtaining physical state data to be evaluated of a pilot to be evaluated in a ground stage on a continuous time, and predicting by a pilot flight stage physical state prediction model to obtain predicted physical state data, including:
acquiring to-be-evaluated physical state data of a pilot to be evaluated in a ground stage on continuous time, and determining to-be-evaluated physical state data in the first n time steps based on the current time on the basis of the acquired to-be-evaluated physical state data of the ground stage on continuous time;
and taking the determined physical state data to be evaluated in the first n time steps as input of a pilot flight stage physical state prediction model to obtain predicted physical state data in the next m time steps.
In one possible embodiment, the method includes identifying predicted physical state data by a pilot health value assessment model, obtaining a predicted health value for a future time period, and assessing a health flight duration based on the predicted health value for the future time period, including:
taking the predicted physical state data of m time steps in the future as the input of a pilot health value evaluation model to obtain predicted health values of m time steps in the future;
Based on the predicted health values of the m time steps in the future, performing curve fitting by adopting a spline interpolation method to obtain a time step health value curve;
judging whether a point smaller than a preset health threshold exists in the time step health numerical curve, if so, taking the time corresponding to the first point smaller than the preset health threshold as a first target time, otherwise, determining that the health flight duration is the duration formed by m time steps in the future;
and determining the healthy flight duration to be the duration consisting of k-1 time steps in the future based on the time step k of the first target time.
In one possible embodiment, the method further comprises:
judging whether a point with a slope larger than a preset slope threshold exists in the time step health numerical curve, if so, taking the time corresponding to the point with the first slope larger than the preset slope threshold as a first target time, otherwise, determining that the health flight duration is the duration formed by m time steps in the future;
and determining the healthy flight duration to be the duration consisting of k-1 time steps in the future based on the time step k of the first target time.
In one possible embodiment, the method further comprises:
and when no point smaller than the preset health threshold value and no point with the slope larger than the preset slope threshold value exist in the time step health value curve, continuously acquiring the predicted health value of the future period, and evaluating the health flight duration according to the predicted health value of the future time.
As shown in fig. 2, an embodiment of the present invention provides a system for evaluating healthy flight time of a pilot in civil aviation, including: the system comprises a pilot physical state data acquisition module 1, a pilot flight phase physical state prediction module 2, a flight phase health value evolution prediction module 3 and a health flight duration evaluation module 4.
The pilot physical state data acquisition module 1 is used for acquiring physical state data and health state data of a civil aviation pilot in continuous time and constructing a pilot physical state prediction database according to the physical state dataDTBuilding a health value evaluation database according to the physical state data and the health state dataDR。
The health related data actively detected by the civil aviation pilot in the ground work and flight task can be collected. It should be noted that, in view of the many complex situations that civil pilots may face in flight, there are limitations in relying on only a single type of data for health analysis due to their occupational specificity. For this reason, the present embodiment starts with relevant indicators that affect the health of the pilot, comprehensively analyzing multiple types of data that can be actively detected in ground work and flight tasks. The comprehensive analysis is performed in three dimensions, namely pilot basic state data, fatigue state data and physiological index data, so as to comprehensively depict the physical condition of the civil aviation pilot.
It should be further elaborated that the acquisition of pilot basic state data, fatigue state data and physiological index data in three dimensions is achieved by the following steps:
the pilot physical state data acquisition module acquires pilot basic state data, acquires the height, weight, age, physical ability and other data of the pilot uniformly through the pilot database, and adopts unique identification codes to represent the data, so that personal privacy of the pilot in civil aviation is protected.
The pilot physical state data acquisition module acquires the physiological index data of the civil aviation pilot during flight, and acquires the physiological index data through an intelligent bracelet/watch worn by the civil aviation pilot in real time, wherein the physiological index data comprises parameters including, but not limited to, heart rate HR, blood oxygen saturation SpO2, body temperature Temp, sleep duration Sleep and the like.
Fatigue state data of the civil aviation pilot during flight is obtained through the pilot physical state data collecting module, high-precision image collecting equipment is deployed in a place where the civil aviation pilot frequently moves, and particularly, ground working areas, flight training scenes and flight executing scenes are obtained to obtain image data of the civil aviation pilot. After preprocessing through geometric transformations such as translation, scaling, rotation, flipping, miscut, and the like, and image enhancement algorithms such as gaussian filtering, facial keypoints are identified from pilot images using a pre-trained pilot facial keypoint identification model (DCNN (Dynamic Convolution Neural Network, deep convolutional neural network) or the like algorithms may be employed) to locate eye positions. The training of the pilot face key point recognition model adopts a migration learning strategy, the weight parameters capable of recognizing the universal face key points are obtained through a pre-training face database, and the weight parameters are migrated to a civil aviation pilot face key point training task, so that the accuracy of key point recognition is improved. And positioning the critical point information to eyes and eyelids of the pilot, calculating the closing degree of the eyes, and judging that the eyes are closed when the closing degree is more than 50%. And pass through PERCLOSThe measurement principle judges fatigue as shown in the following formula:
；
in the method, in the process of the invention,ECfor the number of frames that the eye is closed,TFis total per unit timeFrame number, whenPERCLOSIf the value is greater than the set threshold value, the fatigue is obtained, and the value indicates the degree of fatigue. Based on the latest sleep time data corresponding to the time when the pilot is determined to be tiredSleepWhen the sleeping time is enoughsleep>8h) If the vehicle is still in a fatigue state, judging that the vehicle cannot relieve fatigue through rest, otherwise judging that the vehicle is tiredPL. Finally, fatigue state data of the pilot in unit time are obtained: non-fatiguen_PLFatigue ofPLFatigue that cannot be relieved by restc_PL. It should be noted that only two states, namely, a fatigued state and a non-fatigued state, may be employed.
Building pilot body state prediction database from collected dataDTAnd health value assessment databaseDR. Specifically, the basic state data, the physiological index data and the fatigue state data of the pilot with the same time scale in a longer period are collected for a single individual civil aviation pilot, and corresponding time information (including but not limited to time, time difference, time zone and the like) is recorded at the same time to form a pilot physical state databaseDT. Meanwhile, health numerical data of comprehensive evaluation of physical examination of civil aviation pilots are obtained, three indexes are matched according to time dimension and used as labels of the indexes, and a health numerical evaluation database is formed DR. Both databases are continuously updated with the continuous detection of the acquisition device.
A pilot flight phase physical state prediction module 2 for predicting a pilot physical state databaseDTAnd learning the data in the model to obtain the physical state prediction model of the pilot in the flight phase.
In particular, the predictive database formed by the pilot body state data acquisition module may be based onDTTraining a pilot flight phase physical state prediction model to predict the future of a civil aviation pilotmPhysical state over a time of flight. First, to a databaseDTThe data in the model is subjected to detailed preprocessing, including data cleaning, standardization and the like, so as to ensure the consistency of the data and the robustness of the model. Taking into account direct passage through a databaseDTThe physical state of the pilot is predicted, and the predicted result is difficult to be defined as before-flight and in-flightOr the state between the two, and the physical state of the civil aviation pilot before the flight mission can obviously influence the physical state in the flight process.
For this purpose, civil aviation pilot work processes can be divided into flight phases (take-off, cruising, navigation, landing and possibly emergency situations) in ground phases (performing work such as pre-flight checks, aircraft maintenance, etc.) and in flight training or in flight. And will be DTThe data of the model is divided into data consisting of the ground phase physical state and the physical state data of the continuous flight phase, so that the pilot flight phase physical state prediction model can predict the flight phase physical state and fully consider the influence of the ground phase physical state on the subsequent trend.
Specifically, the preprocessed data is input into the pilot flight phase physical state prediction model for training. During training, the learning effect of the model on the physical state of the pilot is improved through parameter optimization so as to more accurately predict the evolution of the physical state of the pilot in the flight phase.
It should be noted that, before input, the ground phase physical state data and the flight phase physical state data are respectively weighted by introducing a self-attention mechanism. This helps to highlight the key features of the physical state of each stage, so that the model is easier to capture the influence of the key features of the ground stage when learning the trend of the physical state change of the flight stage, and at the same time effectively prevents the key features of the flight stage from being aggregated to the ground stage. After training is completed, judging whether the current state of the pilot is a flight stage, if so, continuously updating the physical state data of the flight stage in real time, and if not, passing through a fixed length by means of an optimized physical state prediction model of the pilot flight stage nThe physical state of the ground stage of the aircraft is compared with the future flying stagemTime steps) of the pilot's physical state. And simultaneously, updating the physical state data of the ground stage in real time. This prediction may provide data on future physical state of the pilot in civil aviation.
Further, a specific method for building a pilot flight phase physical state prediction model is additionally described, and the method comprises the following real-time steps:
database for preprocessing data such as data cleaning and standardizationDTDivided into data consisting of body state data of ground phase and body state data of continuous flight phase, wherein the length of the body state data of ground phase is fixed asnWhile the physical state data of the flight phase ism(determined by the duration of the flight mission).
Taking a gating circulation unit algorithm as an example, the embodiment of the invention trains a physical state prediction model of a pilot in a flight phase. It should be noted that, in the process of achieving the objective of this step, the physical state of the pilot in the ground stage is considered to have a significant effect on the physical state in the flight process, so that certain improvement is made when the pilot flight stage physical state prediction model is trained by the gating cycle unit algorithm.
Specifically, the embodiment of the invention carries out staged self-attention processing on the ground stage physical state data and the flight stage physical state data to finish the weighted attention of the information of different time steps, and is characterized in that the method can respectively aggregate the relevant information of important time steps in the ground stage physical state and the flight stage pilot physical state and highlight the key characteristics of the two stages physical state, so that the influence of the key characteristics of the ground stage can be obtained more easily when the model learns the change trend of the flight stage physical state. At the same time, the ground stage is prevented from aggregating the key features of the flying stage, because the physical state of the flying stage as a subsequent task of the ground stage does not affect the pre-task.
Separate learning databaseDTThe self-attention weight of the physical state of the ground phase and the flight phase is weighted and summed with each time step feature in the input ground phase and flight phase sequence, so as to highlight each phase key feature.
The ground stage physical state self-attention weight expression is as follows:wherein d represents the ground levelThe attention score of a segment is obtained by calculating the dot product of the segment after the body state data of the ground stage are coded and then dividing the dot product by the characteristic dimension after root opening. Similarly, the self-attention weight expression of the physical state in the flight stage is as follows: / >Where f represents the attention score of the flight phase, which is obtained by calculating the dot product of the flight phase body state data after encoding, and then dividing by the feature dimension after rooting.
Calculating the weighted sum of the physical state and the self-attention of each ground stage and the flight stage respectively, wherein the expressions are as follows: xd and xf, the calculation process of which is shown in the following formula:
；
；
in the method, in the process of the invention,、/>representing the respective positions in the ground phase and the flight phaseiTo the positionjIs self-attention weight of +.>、/>Representing the weighted and summed sequences of ground phases and flight phases respectively,irepresenting the position in the sequence. />、/>Representing an original ground phase and a sequence of flight phases,jrepresenting positions in a sequence。
The embodiment of the invention adopts a gating circulating unit algorithm for training. The training process is expressed as: setting time step n and characteristic quantity of each time step(sum of basic state features, physiological index features and fatigue feature quantity), proper number of iterations +.>Batch size->And the like. During training, weights and biases are adjusted by a back-propagation algorithm to minimize the loss function, enabling the model to learn a complex time-dependent relationship of body state changes from ground phase to flight phase. Specifically, the present embodiment time-sequentially +/s-sequences of n pilot ground phase physical state sequence data observed in the past at the present time >The models are sequentially input as shown in the following formula:
；
in the method, in the process of the invention,indicates the hidden state of time step t-1, < >>An input sequence representing time step t, i.e. the above-mentioned self-attention weighted sum,/i>Representing update door->Representing ground stage weight matrix,/->Representing the bias parameters +.>Representing a sigmoid activation function.
Simultaneous computing reset gateThe following formula:
；
in the method, in the process of the invention,representing a weight matrix, +.>Representing the bias parameters.
Calculating new candidate hidden statesThe following formula:
；
in the method, in the process of the invention,representing a weight matrix, +.>Representing the bias parameters +.>Representing a hyperbolic tangent activation function, e represents an element-wise multiplication.
Hidden state at update time step tIs calculated as follows:
；
the calculation process is repeated in sequence until the time sequence data of the physical state of the n pilots are calculated according to the respective corresponding time sequence, and the hidden state of the first time step of the flight phase is output. And predicts the physical state of the civil aviation pilot at this moment +.>The following formula:
；
in the method, in the process of the invention,is a fully connected layer that maps hidden states to predicted outputs. Subsequent continuation of the self-attention weighted flight phase dataxfAs a new observation, the procedure is repeated until all +. >To->And predicting the physical state of the pilot in the future flight phase, so as to form a pilot flight phase physical state prediction model. The improvement mode can be used for easily learning the influence of the physical state key features of the ground stage on the physical state key features of the flight stage while highlighting the physical state key features of the flight stage, so that the aim of improving the prediction accuracy of the physical state change trend of the pilot in the flight stage is fulfilled.
The physical state prediction of the pilot flight stage is that the physical state prediction model of the pilot flight stage is trained, the physical state data to be evaluated of the ground stage of the pilot to be evaluated in continuous time is input, the physical state data to be evaluated of the ground stage of the pilot to be evaluated in continuous time is output, the time step flight stage is output, the time step predicted value is added into an input sequence to form a new n-order input sequence, and the physical state data of m time step flight stages in the future are predicted in a recursion iteration mode
A flight phase health value evolution prediction module 3 for evaluating the health value databaseDRLearning is carried out on the data in the model to obtain a pilot health numerical value assessment model, and the flight phase health numerical evolution prediction is realized based on the pilot health numerical value assessment model.
First, a health value evaluation database is adoptedDRTraining is performed to obtain the construction of the pilot health numerical assessment model. In view of the difficulty in acquiring the health numerical value label of part of the flight stage, a semi-supervised training method is adopted for model optimization. Based on the well-trained model, the future flight phase in the pilot flight phase physical state prediction module is calculatedmTime steps) are used as input, and the future flight phase of the pilot is obtained through a health numerical evaluation modelmIndividual time steps). And finally, approximating the future integral evolution form of the health value by adopting a curve fitting mode, and finishing the evolution prediction of the health value in the flight stage.
For this purpose, the present embodiment employs a BP (back propagation) neural network algorithm. Specifically, the weights and biases in the neural network are randomly initialized first, then the physical state of the pilot is input into the neural network, and the output of each layer is calculated as shown in the following formula:
；
in the method, in the process of the invention,is the firstlLayer to the firstl-1) a weight matrix of layers; />Is the firstlA bias vector for the layer; />To activate the function. Will godAnd comparing the output value through the network with the health numerical value tag through a mean square error function to obtain an error. The error contribution for each layer is calculated using the chain law and the weights and offsets are updated. In view of a certain difficulty in acquiring the health numerical value labels of part of the flight phases, the labels cannot be matched with all the physical state data in the time dimension, and a semi-supervised training method is adopted for model optimization.
Based on the well-trained model, the predicted future flying stage is calculatedmTime steps) are used as input, and the future flight phase of the pilot is obtained through a health numerical evaluation modelmIndividual time steps). Finally, the embodiment adopts a spline interpolation method to complete predictionmAnd fitting a time step health value curve, approaching to the future integral evolution form of the health value, and finishing the evolution prediction of the health value in the flight stage.
And the healthy flight duration evaluation module 4 is used for evaluating the healthy flight duration according to the flight phase health value evolution prediction.
Evaluation of a database by health values taking into account differences in individual health criteria of pilotsDRThe mean minus the standard deviation is set as the threshold for health imbalance. Combining the prediction of future pilot health value evolution, if judgingkA time step (in whichk<m) Observing that the health value is below the threshold and continues to be below the threshold for a longer period of time, it can be inferred thatkTime-1 is the optimal flight time that the pilot can be qualified without affecting health. Considering that pilots are subjected to high pressure conditions during the flight phase, the evolution of their health values may be subject to dips, leading to erroneous decisions and potential safety risks.
Thus, even if the health value is not yet below the threshold, if at the firstkThe health value observed at each time step can be considered to be the best time of flight that the pilot can be qualified before this time point if the slope of the value evolution suddenly drops at this point. If none of the above conditions are met, it is necessary to continuously observe the evolution of the health value during the flight phase. This isThe judgment basis is helpful for identifying potential risks of health in advance, and the health and safety of pilots are guaranteed.
The embodiment of the invention designs an evaluation model of the healthy flight duration of the civil aviation pilot from the viewpoint of the health of the civil aviation pilot. On the one hand, the model firstly collects physical state data of a pilot in historical work, and the process fully considers that the civil aviation pilot can face various complex situations in flight due to the occupational specificity of the civil aviation pilot, and the problem that a single type of data is not comprehensive in use for health analysis exists. And comprehensively analyzes the data which can be actively detected in each working stage of the pilot and has influence on the health of the pilot, so that a richer information visual angle can be provided in the subsequent modeling, and the characteristics and behaviors related to the health of the pilot can be more comprehensively and accurately mastered. On the other hand, potential relation between the health value and the flight duration of the future flight phase of the pilot can be determined through the physical state prediction model and the health value evaluation model of the flight phase of the pilot, and the trend of the change of the health value along with the flight duration is excavated. The optimal flight time which can be ensured to be qualified under the condition of not affecting the health can be obtained through analysis.
The method comprises the steps of fully analyzing and directly predicting the physical state of a pilot in the construction of a physical state prediction model of the pilot flight stage, and solving the problems that the predicted result is the state before flight, in the flight process or between the two, and considering the influence of the physical state of the pilot ground stage on the physical state of the flight stage. The self-attention mechanism is respectively used for the physical state data of the ground stage and the physical state data of the flight stage, and the information of different time steps is weighted and focused, so that the influence of the key characteristics of the model can be highlighted when the model learns the physical state change trend of the flight stage, and the influence of the key characteristics of the ground stage on the model can be received more easily. And meanwhile, the key characteristics of the ground stage and the aggregation flight stage are prevented, so that the health flight duration of subsequent evaluation is more reasonable.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (10)
1. A method for evaluating the healthy flight time of a pilot in civil aviation, comprising:
acquiring physical state data and health state data of a civil aviation pilot in continuous time, constructing a pilot physical state prediction database DT according to the physical state data, and constructing a health value evaluation database DR according to the physical state data and the health state data;
taking the pilot physical state prediction database DT as a data support to train a pilot flight phase physical state prediction model; training a pilot health value evaluation model by taking the health value evaluation database DR as data support;
acquiring to-be-evaluated physical state data of a pilot to be evaluated in a ground stage in continuous time, and predicting through a pilot flight stage physical state prediction model to obtain predicted physical state data;
and identifying the predicted physical state data through a pilot health value evaluation model to obtain a predicted health value of a future period, and evaluating the health flight duration according to the predicted health value of the future time.
2. The method of assessing a health time of flight of a pilot of a civil aircraft of claim 1, wherein the physical state data includes pilot base state data, fatigue state data, and physiological index data;
Wherein the pilot basic state data includes height, weight, age and physical performance indicators;
the fatigue state data comprise that a pilot is in a general fatigue state, a fatigue state which cannot be relieved through rest or a flight is in a normal state;
the physiological index data includes heart rate, blood oxygen saturation, body temperature, and/or sleep duration.
3. The method for evaluating the healthy flight time of a civil aircraft pilot according to claim 2, wherein the method for acquiring the fatigue state data comprises:
and acquiring a plurality of frames of pilot face images, positioning pilot eye positions through a pilot face key point recognition model trained by transfer learning, and determining the fatigue state type of the pilot by combining a PERCLOS measurement principle and latest sleep duration data.
4. A method of assessing a healthy flight time of a pilot for civil aviation as claimed in claim 3 wherein determining the type of fatigue status of the pilot in combination with PERCLOS measurement principles and recent sleep duration data comprises:
based on the eye position of the pilot, determining eye opening and closing information corresponding to each frame of pilot face image; the eye opening and closing information comprises eye opening or eye closing;
According to eye opening and closing information corresponding to all pilot face images, the fatigue index is determined by combining a PERCLOS measurement principle to be:
;
wherein,PERCLOSthe fatigue index is indicated as a function of the fatigue index,ECthe number of frames representing the eye closure,TFrepresenting the total number of frames of the acquired pilot face image;
judging whether the fatigue index is larger than a preset fatigue threshold value, if so, determining that the fatigue state type is that the pilot is in a fatigue state, otherwise, determining that the fatigue state type is that the pilot is in a normal state;
judging whether the sleep duration data is larger than a preset sleep threshold value, if so, determining that the fatigue state type is a fatigue state which can not be relieved by rest, otherwise, determining that the fatigue state type is a general fatigue state.
5. The method for evaluating the healthy flight time of a pilot for civil aviation according to claim 1, characterized in that training a pilot's flight phase physical state prediction model with the pilot physical state prediction database DT as a data support comprises:
dividing physical state data of a civil aviation pilot in continuous time into ground-stage physical state data with a fixed length of n and flight-stage physical state data with a fixed length of m by taking the pilot physical state prediction database DT as a data support;
The method comprises the steps of constructing a first initial prediction model through a deep learning algorithm, taking ground stage body state data with a fixed length of n as actual input data of the first initial prediction model, taking flight stage body state data with a fixed length of m as expected prediction data of the first initial prediction model, and training the first initial prediction model to obtain a pilot flight stage body state prediction model.
6. The method for assessing a healthy flight time of a pilot for civil aviation according to claim 5, wherein training a pilot health value assessment model with the health value assessment database DR as data support comprises:
taking the health value evaluation database DR as a data support, and extracting the physical state data of the flight stage with the fixed length of m and the physical state data of the flight stage with the fixed length of m, which correspond to the physical state data of the flight stage;
and constructing a second initial prediction model through a deep learning algorithm, taking the physical state data of the flight stage with the fixed length of m as actual input data of the second initial prediction model, taking the physical state data of the flight stage with the fixed length of m corresponding to the physical state data of the flight stage as expected prediction data of the second initial prediction model, and training the second initial prediction model to obtain a pilot health numerical value assessment model.
7. The method for assessing a healthy flight time of a pilot for civil aviation according to claim 6, wherein obtaining the physical state data to be assessed of the pilot to be assessed at ground stage over continuous time and predicting by a pilot flight stage physical state prediction model to obtain predicted physical state data comprises:
acquiring to-be-evaluated physical state data of a pilot to be evaluated in a ground stage on continuous time, and determining to-be-evaluated physical state data on the first n time steps based on the current time on the basis of the obtained to-be-evaluated physical state data on continuous time;
and taking the determined physical state data to be evaluated in the first n time steps as input of a pilot flight stage physical state prediction model to obtain predicted physical state data in the next m time steps.
8. The method of claim 7, wherein the step of identifying the predicted physical state data by the pilot health value assessment model to obtain a predicted health value for the future period and assessing the healthy flight duration based on the predicted health value for the future period comprises:
Taking the predicted physical state data of m time steps in the future as the input of a pilot health value evaluation model to obtain predicted health values of m time steps in the future;
based on the predicted health values of the m time steps in the future, performing curve fitting by adopting a spline interpolation method to obtain a time step health value curve;
judging whether a point smaller than a preset health threshold exists in the time step health numerical curve, if so, taking the time corresponding to the first point smaller than the preset health threshold as a first target time, otherwise, determining that the health flight duration is the duration formed by m time steps in the future;
and determining the healthy flight duration to be the duration consisting of k-1 time steps in the future based on the time step k of the first target time.
9. The method of assessing a healthy flight time of a pilot for a civil aircraft of claim 8, further comprising:
judging whether a point with a slope larger than a preset slope threshold exists in the time step health numerical curve, if so, taking the time corresponding to the point with the first slope larger than the preset slope threshold as a first target time, otherwise, determining that the health flight duration is the duration formed by m time steps in the future;
And determining the healthy flight duration to be the duration consisting of k-1 time steps in the future based on the time step k of the first target time.
10. The method of assessing a healthy flight time of a pilot for a civil aircraft of claim 8, further comprising:
and when no point smaller than the preset health threshold value and no point with the slope larger than the preset slope threshold value exist in the time step health value curve, continuously acquiring the predicted health value of the future period, and evaluating the health flight duration according to the predicted health value of the future time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410173526.3A CN117727449B (en) | 2024-02-07 | 2024-02-07 | Evaluation method for healthy flight time of civil aviation pilot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410173526.3A CN117727449B (en) | 2024-02-07 | 2024-02-07 | Evaluation method for healthy flight time of civil aviation pilot |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117727449A true CN117727449A (en) | 2024-03-19 |
| CN117727449B CN117727449B (en) | 2024-05-17 |
Family
ID=90207322
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410173526.3A Active CN117727449B (en) | 2024-02-07 | 2024-02-07 | Evaluation method for healthy flight time of civil aviation pilot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117727449B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104680852A (en) * | 2013-11-26 | 2015-06-03 | 空中客车运营简化股份公司 | Mission management system and method of aircraft |
| CN109528157A (en) * | 2017-09-22 | 2019-03-29 | 极光飞行科学公司 | System and method for monitoring pilot's health |
| CN110406512A (en) * | 2019-07-04 | 2019-11-05 | 北京长城华冠汽车技术开发有限公司 | The body health monitoring devices and system of automobile |
| CN110427988A (en) * | 2019-07-17 | 2019-11-08 | 陕西千山航空电子有限责任公司 | A kind of Airborne Flight Parameter data Gernral Check-up method based on deep learning |
| CN110623680A (en) * | 2019-08-30 | 2019-12-31 | 中国民用航空飞行学院 | Method for testing psychological health of civil aviation flight trainees |
| CN112766576A (en) * | 2021-01-22 | 2021-05-07 | 无锡市第五人民医院 | Parameter result prediction method and system based on database |
| CN114010159A (en) * | 2021-11-10 | 2022-02-08 | 中防通用河北电信技术有限公司 | Prediction system for predicting physical health state of user |
| CN114642404A (en) * | 2022-03-17 | 2022-06-21 | 中国民用航空飞行学院 | Pilot physical fitness health monitoring equipment and operation method thereof |
| CN114912829A (en) * | 2022-06-02 | 2022-08-16 | 中国民用航空飞行学院 | Method and system for evaluating flight suitability of pilots in plateau |
| CN116196002A (en) * | 2023-04-28 | 2023-06-02 | 中国人民解放军空军特色医学中心 | Method and system for evaluating physiological stress of pilot in air |
| CN116596385A (en) * | 2023-05-17 | 2023-08-15 | 浙江省交通运输科学研究院 | Pilot navigability assessment method and device, storage medium and electronic equipment |
-
2024
- 2024-02-07 CN CN202410173526.3A patent/CN117727449B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104680852A (en) * | 2013-11-26 | 2015-06-03 | 空中客车运营简化股份公司 | Mission management system and method of aircraft |
| CN109528157A (en) * | 2017-09-22 | 2019-03-29 | 极光飞行科学公司 | System and method for monitoring pilot's health |
| CN110406512A (en) * | 2019-07-04 | 2019-11-05 | 北京长城华冠汽车技术开发有限公司 | The body health monitoring devices and system of automobile |
| CN110427988A (en) * | 2019-07-17 | 2019-11-08 | 陕西千山航空电子有限责任公司 | A kind of Airborne Flight Parameter data Gernral Check-up method based on deep learning |
| CN110623680A (en) * | 2019-08-30 | 2019-12-31 | 中国民用航空飞行学院 | Method for testing psychological health of civil aviation flight trainees |
| CN112766576A (en) * | 2021-01-22 | 2021-05-07 | 无锡市第五人民医院 | Parameter result prediction method and system based on database |
| CN114010159A (en) * | 2021-11-10 | 2022-02-08 | 中防通用河北电信技术有限公司 | Prediction system for predicting physical health state of user |
| CN114642404A (en) * | 2022-03-17 | 2022-06-21 | 中国民用航空飞行学院 | Pilot physical fitness health monitoring equipment and operation method thereof |
| CN114912829A (en) * | 2022-06-02 | 2022-08-16 | 中国民用航空飞行学院 | Method and system for evaluating flight suitability of pilots in plateau |
| CN116196002A (en) * | 2023-04-28 | 2023-06-02 | 中国人民解放军空军特色医学中心 | Method and system for evaluating physiological stress of pilot in air |
| CN116596385A (en) * | 2023-05-17 | 2023-08-15 | 浙江省交通运输科学研究院 | Pilot navigability assessment method and device, storage medium and electronic equipment |
Non-Patent Citations (3)
| Title |
|---|
| 唐天雄: ""飞行疲劳与风险管理的研究"", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》, 15 May 2022 (2022-05-15), pages 026 - 57 * |
| 沈锋等: ""神经网络在飞行疲劳检测中的应用综述"", 《软件导刊》, 15 March 2022 (2022-03-15), pages 244 - 248 * |
| 邹琳: ""民航飞行员飞行疲劳体征的判别与安全对策综述"", 《才智》, 5 July 2011 (2011-07-05), pages 325 - 327 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117727449B (en) | 2024-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cui et al. | The empathic framework for task learning from implicit human feedback | |
| Brown-Brandl et al. | Evaluating modelling techniques for cattle heat stress prediction | |
| Hamdan et al. | Adaptive neuro-fuzzy inference system (ANFIS) in modelling breast cancer survival | |
| CN111462191B (en) | Non-local filter unsupervised optical flow estimation method based on deep learning | |
| Begum et al. | Classification of physiological signals for wheel loader operators using multi-scale entropy analysis and case-based reasoning | |
| Chitra et al. | Heart attack prediction system using fuzzy C means classifier | |
| CN115063676A (en) | Ship target classification method based on AIS data | |
| Mehmood et al. | Systematic framework to predict early-stage liver carcinoma using hybrid of feature selection techniques and regression techniques | |
| Suzdaltsev et al. | Fuzzy rules formation for the construction of the predictive diagnostics expert system | |
| CN112560948A (en) | Eye fundus map classification method and imaging method under data deviation | |
| Dhindsa et al. | Grading prenatal hydronephrosis from ultrasound imaging using deep convolutional neural networks | |
| Langeroudi et al. | FD-LSTM: A fuzzy LSTM model for chaotic time-series prediction | |
| Walia et al. | A survey on applications of adaptive neuro fuzzy inference system | |
| CN117727449B (en) | Evaluation method for healthy flight time of civil aviation pilot | |
| Chen et al. | A situation awareness assessment method based on fuzzy cognitive maps | |
| Sundari et al. | Prediction of Activity pattern mining for Neurological disease using Convolution Neural Network | |
| Choudhary et al. | Glaucoma detection using cross validation algorithm | |
| Li et al. | Glaucoma detection: Joint segmentation and classification framework via deep ensemble network | |
| Nam et al. | Selective prediction with long short-term memory using unit-wise batch standardization for time series health data sets: algorithm development and validation | |
| CN116269212A (en) | Multi-mode sleep stage prediction method based on deep learning | |
| CN113343774B (en) | Fine-grained engineering mechanics diving action simulation and evaluation method | |
| Elakkiya et al. | A comparative analysis of pretrained and transfer-learning model for automatic diagnosis of glaucoma | |
| Agustin et al. | Optimization Convolutional Neural Network for Classification Diabetic Retinopathy Severity | |
| Rastgoo | Driver stress level detection based on multimodal measurements | |
| Anam et al. | Classification Model for Diabetes Mellitus Diagnosis based on K-Means Clustering Algorithm Optimized with Bat Algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |