CN118210292A - SR20 type airplane fault detection method and system based on visualization technology - Google Patents

Abstract

The invention discloses a SR20 type aircraft fault detection method and system based on a visualization technology, which relate to the technical field of fault detection, wherein the method comprises the following steps: acquiring historical flight state data and historical weather data of SR20 type aircraft at different flight heights; and real-time flight status data at the current flight altitude, real-time weather data at the current flight altitude; training a regression network model for predicting SR20 type aircraft flight state data; inputting real-time weather data at the current flight altitude into a regression network model to obtain SR20 type aircraft flight state data predicted based on the current flight altitude, substituting the SR20 type aircraft flight state data and the real-time flight state data at the current flight altitude into a state error calculation strategy, and calculating a state error in the SR20 type aircraft real-time flight state data; and carrying out fault early warning on the abnormal state and feeding back to the pilot. The safety of the SR20 type airplane in flying can be further improved.

Description

SR20 type airplane fault detection method and system based on visualization technology

Technical Field

The invention relates to the technical field of fault detection, in particular to a SR20 type aircraft fault detection method and system based on a visualization technology.

Background

In the field of aviation, the safety and reliability of an aircraft are critical, and for the aspect of aircraft fault detection, visualization techniques are widely used to help engineers and technicians to discover and diagnose aircraft faults more quickly and accurately. For example, for an SR20 type airplane, the state data, sensor information and the like of all parts of the airplane can be checked through a visualization technology, so that potential faults can be found in time, and the running safety and reliability of the airplane are improved. However, the fault detection of the aircraft involves a plurality of data sources and systems, the data sources need to be integrated and analyzed, and the fault detection method needs to be correspondingly adjusted for different flying heights when the aircraft flies.

The patent in China with the publication number of CN117092982A discloses an aircraft fault detection system and a working method thereof, and relates to the technical field of fault detection. The working method of the system comprises the steps of carrying out data standardization processing and principal component analysis on historical detection data and current detection data of the flight control system, determining an adaptive threshold value for evaluating whether the sensor fails according to the principal component analysis, and determining whether the sensor fails according to the magnitude relation between a second threshold value determined by the current detection data and the adaptive threshold value. The system and the working method thereof disclosed by the invention can timely detect the sensor faults, so that maintenance personnel can timely confirm and maintain the faults, and the safe and stable operation of the aircraft is ensured.

The chinese patent application publication No. CN109557896a discloses a system and method for aircraft fault detection. An aircraft fault detection system comprising: at least one aircraft data recording device configured to capture parametric flight data from at least one aircraft subsystem; and an aircraft controller coupled to the data recording device. The controller is configured to group parametric flight data from the at least one aircraft subsystem into a plurality of test states, one or more of the plurality of states being different from the other test states, generate at least one test transition matrix based on the plurality of test states, and determine an abnormal behavior of the at least one aircraft subsystem based on the at least one test transition matrix, and predict a fault within the at least one aircraft subsystem based on the abnormal behavior of the at least one aircraft subsystem determined from the at least one test transition matrix.

The problems presented in the background art exist in the above patents: the flight state fault monitoring method for the aircraft under different flying heights and weather conditions is needed because the flight state of the aircraft under different flying heights is not considered to be different, the generated data are different, and the weather of the positions of the aircraft under different flying heights is also different. In order to solve the problems, the invention provides an SR20 type airplane fault detection method and system based on a visualization technology.

Disclosure of Invention

Aiming at the defects of the prior art, the main purpose of the invention is to provide a SR20 type aircraft fault detection method and system based on a visualization technology, which can effectively solve the problems in the background technology. The specific technical scheme of the invention is as follows:

The SR20 type airplane fault detection method based on the visualization technology comprises the following steps:

S1, acquiring multiple flying heights of an SR20 type airplane, and acquiring historical flight state data of the SR20 type airplane at different flying heights and historical weather data at different flying heights based on the multiple flying heights;

S2, acquiring real-time flight state data of the SR20 type aircraft at the current flight altitude and real-time weather data of the SR20 type aircraft at the current flight altitude;

S3, training a regression network model for predicting the SR20 type aircraft flight state data based on the flight altitude, the historical flight state data and the historical weather data;

s4, inputting real-time weather data under the current flight altitude into the regression network model based on the regression network model to obtain SR20 type aircraft flight state data predicted under the current flight altitude;

S5, substituting the SR20 type airplane flight state data predicted based on the current flight altitude and the real-time flight state data of the SR20 type airplane under the current flight altitude into a state error calculation strategy, and calculating the state error in the SR20 type airplane real-time flight state data;

s6, arranging the state errors in a descending order, presetting a state error threshold value, carrying out fault early warning on an abnormal state corresponding to the state error larger than the state error threshold value in the real-time flight state data, and feeding back abnormal component data corresponding to the abnormal state to a pilot.

Specifically, the flight state data comprise engine parameters, oil pressure, engine temperature, flight attitude and flight speed; the weather data includes wind speed, wind direction, air humidity, visibility, temperature.

Specifically, the specific content of S1 further includes:

And establishing a flight data set A _h＝{h,W_h,S_h when the flight height is h, wherein W _h is historical weather data when the flight height is h, S _h is historical flight state data when the flight height is h, respectively cleaning the flight data sets at different flight heights, checking whether missing values exist in the historical weather data and the historical flight state data, deleting the flight data sets at the corresponding flight heights containing the missing values if the missing values exist, and taking the flight data sets at all flight heights after the data cleaning as a model data set.

Specifically, the S3 further includes the following specific contents: dividing a model data set into a model training set and a model testing set, and constructing a regression network model; taking the flying height in the flying data set under all flying heights in the model training set, the historical wind speed, the historical wind direction, the historical air humidity, the historical visibility and the historical temperature in the historical weather data as inputs of a regression network model, taking the historical engine parameters, the historical oil pressure, the historical engine temperature, the historical flying posture and the historical flying speed in the historical flying state data in the flying data set under all flying heights in the model training set as outputs of the regression network model, training the model to obtain an initial regression network model, and evaluating the model effect of the initial regression network model by utilizing a mean square error algorithm and presetting an evaluation value. And screening a corresponding initial regression network model with the mean square error smaller than or equal to a preset evaluation value as a regression network model for predicting the SR20 type aircraft flight state data.

Specifically, the specific steps of obtaining the initial regression network model include:

S31, constructing a regression network model comprising an input layer, a hidden layer and an output layer, wherein the input layer comprises 6 input nodes, and the fly height, the wind speed, the wind direction, the air humidity, the visibility and the temperature are respectively input through the 6 input nodes; the output layer comprises 5 output nodes, and each output node respectively corresponds to and outputs an engine parameter predicted value, an oil pressure predicted value, an engine temperature predicted value, a flight attitude predicted value and a flight speed predicted value;

S32, the hidden layer output formula is h=σ (W ₁×X+b₁), where H is the hidden layer output, W ₁ is the hidden layer weight, b ₁ is the bias, σ is the activation function, where x= [ X ₁,X₂,X₃,X₄,X₅,X₆ ] is the input feature, X ₁ is the input flying height, X ₂ is the input wind speed, X ₃ is the input wind direction, X ₄ is the input air humidity, X ₅ is the input visibility, and X ₆ is the input temperature;

S33, the predicted value output by the output layer is predicted flight state data Y＝[W₂₁,W₂₂,W₂₃,W₂₄,W₂₅]×H+[b₂₁,b₂₂,b₂₃,b₂₄,b₂₅],, wherein Y＝[Y₁,Y₂,Y₃,Y₄,Y₅],Y₁、Y₂、Y₃、Y₄、Y₅ is an engine parameter predicted value, an oil pressure predicted value, an engine temperature predicted value, a flight attitude predicted value and a flight speed predicted value output by 5 output nodes of the output layer respectively, W ₂₁、W₂₂、W₂₃、W₂₄、W₂₅ is the weight from the hidden layer to 5 output nodes on the output layer respectively, and b ₂₁、b₂₂、b₂₃、b₂₄、b₂₅ is the bias of 5 output nodes on the output layer respectively.

Specifically, the specific steps of evaluating the model effect of the initial regression network model by using the mean square error algorithm are as follows:

s34, the mean square error algorithm is Wherein G is the number of flight data sets A _h in the model test set C, S _h is historical flight state data in the model test set C, and Y is flight state data predicted by an initial regression network model;

S35, presetting an evaluation value T, and taking the corresponding initial regression network model as a regression network model for predicting the SR20 type aircraft flight state data when the MSE is less than or equal to T.

Specifically, the state error calculation strategy includes: acquiring the current flight altitude of the SR20 type aircraft, real-time weather data and real-time flight state data at the current flight altitude, inputting the current flight altitude and the real-time weather data at the current flight altitude into a trained regression network model for predicting the SR20 type aircraft flight state data to obtain predicted flight state data Y＝[Y₁,Y₂,Y₃,Y₄,Y₅],Y₁、Y₂、Y₃、Y₄、Y₅ which are respectively engine parameter predicted values, oil pressure predicted values, engine temperature predicted values, flight attitude predicted values and flight speed predicted values, wherein the acquired real-time engine parameters, real-time oil pressure, real-time engine temperature, real-time flight attitude and real-time flight speed are respectively K ₁、K₂、K₃、K₄、K₅, respectively calculating an engine parameter state error D1, an oil pressure state error D2, an engine temperature state error D3, a flight attitude state error D4 and a flight speed state error D5, respectively presetting an engine parameter state error threshold U1, an oil pressure state error threshold U2, an engine temperature state error threshold U3, a flight attitude state error threshold U4 and a flight speed state error threshold U5, and a state error calculation formula beingAnd when Di is larger than Ui, carrying out abnormal state fault early warning on the ith data in the real-time flight state data corresponding to the ith state error, and feeding back abnormal component data corresponding to the abnormal state to the pilot.

The SR20 type aircraft fault detection system based on the visualization technology is realized based on the SR20 type aircraft fault detection method based on the visualization technology, and comprises the following modules:

the data acquisition module is used for acquiring the multiple flying heights of the SR20 type aircraft, and acquiring historical flight state data of the SR20 type aircraft under different flying heights and historical weather data under different flying heights based on the multiple flying heights; acquiring real-time flight state data of an SR20 type airplane at the current flight altitude and real-time weather data of the SR20 type airplane at the current flight altitude;

The regression network model building module is used for training a regression network model for predicting the SR20 type aircraft flight state data based on the flight altitude, the historical flight state data and the historical weather data;

The model prediction module is used for inputting real-time weather data under the current flight altitude into the regression network model based on the regression network model to obtain SR20 type aircraft flight state data predicted under the current flight altitude;

The state error calculation module is used for substituting the SR20 type airplane flight state data predicted based on the current flight altitude and the real-time flight state data of the SR20 type airplane at the current flight altitude into a state error calculation strategy to calculate the state error in the SR20 type airplane real-time flight state data;

The fault early warning module is used for arranging the state errors in a descending order, presetting a state error threshold value, carrying out fault early warning on an abnormal state corresponding to the state error larger than the state error threshold value in the real-time flight state data, and feeding back abnormal component data corresponding to the abnormal state to a pilot;

and the control module is used for controlling the operation of each module.

Compared with the prior art, the invention has the following beneficial effects:

The invention discloses a SR20 type aircraft fault detection method and system based on a visualization technology, which relate to the technical field of fault detection, wherein the method comprises the following steps: acquiring historical flight state data of an SR20 type airplane at different flight heights and historical weather data at different flight heights; and real-time flight status data at the current flight altitude, real-time weather data at the current flight altitude; training a regression network model for predicting SR20 type aircraft flight state data; inputting real-time weather data at the current flight altitude into a regression network model to obtain SR20 type aircraft flight state data predicted based on the current flight altitude, substituting the SR20 type aircraft flight state data and the real-time flight state data at the current flight altitude into a state error calculation strategy, and calculating a state error in the SR20 type aircraft real-time flight state data; and carrying out fault early warning on the abnormal state, and feeding corresponding abnormal part data back to the pilot. The safety of the SR20 type airplane in flying can be further improved.

Drawings

FIG. 1 is a workflow diagram of an SR20 type aircraft fault detection method based on visualization technology of the present invention;

fig. 2 is a schematic block diagram of an SR20 type aircraft fault detection system based on the visualization technology according to the present invention.

Detailed Description

In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.

Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.

Example 1

The embodiment provides an SR20 type aircraft fault detection method based on a visualization technology, and the specific scheme is that, as shown in fig. 1, the SR20 type aircraft fault detection method based on the visualization technology includes:

S1, acquiring multiple flying heights of an SR20 type airplane, and acquiring historical flight state data of the SR20 type airplane at different flying heights and historical weather data at different flying heights based on the multiple flying heights;

S2, acquiring real-time flight state data of the SR20 type aircraft at the current flight altitude and real-time weather data of the SR20 type aircraft at the current flight altitude;

S3, training a regression network model for predicting the SR20 type aircraft flight state data based on the flight altitude, the historical flight state data and the historical weather data;

s4, inputting real-time weather data under the current flight altitude into the regression network model based on the regression network model to obtain SR20 type aircraft flight state data predicted under the current flight altitude;

S5, substituting the SR20 type airplane flight state data predicted based on the current flight altitude and the real-time flight state data of the SR20 type airplane under the current flight altitude into a state error calculation strategy, and calculating the state error in the SR20 type airplane real-time flight state data;

s6, arranging the state errors in a descending order, presetting a state error threshold value, carrying out fault early warning on an abnormal state corresponding to the state error larger than the state error threshold value in the real-time flight state data, and feeding back abnormal component data corresponding to the abnormal state to a pilot.

In this embodiment, the flight status data includes an engine parameter, an oil pressure, an engine temperature, a flight attitude, and a flight speed; the weather data includes wind speed, wind direction, air humidity, visibility, temperature.

In this embodiment, the specific content of S1 further includes:

And establishing a flight data set A _h＝{h,W_h,S_h when the flight height is h, wherein W _h is historical weather data when the flight height is h, S _h is historical flight state data when the flight height is h, respectively cleaning the flight data sets at different flight heights, checking whether missing values exist in the historical weather data and the historical flight state data, deleting the flight data sets at the corresponding flight heights containing the missing values if the missing values exist, and taking the flight data sets at all flight heights after the data cleaning as a model data set.

In this embodiment, the S3 further includes the following specific contents: dividing a model data set into a model training set and a model testing set, and constructing a regression network model; taking the flying height in the flying data set under all flying heights in the model training set, the historical wind speed, the historical wind direction, the historical air humidity, the historical visibility and the historical temperature in the historical weather data as inputs of a regression network model, taking the historical engine parameters, the historical oil pressure, the historical engine temperature, the historical flying posture and the historical flying speed in the historical flying state data in the flying data set under all flying heights in the model training set as outputs of the regression network model, training the model to obtain an initial regression network model, and evaluating the model effect of the initial regression network model by utilizing a mean square error algorithm and presetting an evaluation value. And screening a corresponding initial regression network model with the mean square error smaller than or equal to a preset evaluation value as a regression network model for predicting the SR20 type aircraft flight state data.

In this embodiment, the specific step of obtaining the initial regression network model includes:

S31, constructing a regression network model comprising an input layer, a hidden layer and an output layer, wherein the input layer comprises 6 input nodes, and the fly height, the wind speed, the wind direction, the air humidity, the visibility and the temperature are respectively input through the 6 input nodes; the output layer comprises 5 output nodes, and each output node respectively corresponds to and outputs an engine parameter predicted value, an oil pressure predicted value, an engine temperature predicted value, a flight attitude predicted value and a flight speed predicted value;

S32, the hidden layer output formula is h=σ (W ₁×X+b₁), where H is the hidden layer output, W ₁ is the hidden layer weight, b ₁ is the bias, σ is the activation function, where x= [ X ₁,X₂,X₃,X₄,X₅,X₆ ] is the input feature, X ₁ is the input flying height, X ₂ is the input wind speed, X ₃ is the input wind direction, X ₄ is the input air humidity, X ₅ is the input visibility, and X ₆ is the input temperature;

S33, the predicted value output by the output layer is predicted flight state data Y＝[W₂₁,W₂₂,W₂₃,W₂₄,W₂₅]×H+[b₂₁,b₂₂,b₂₃,b₂₄,b₂₅],, wherein Y＝[Y₁,Y₂,Y₃,Y₄,Y₅],Y₁、Y₂、Y₃、Y₄、Y₅ is an engine parameter predicted value, an oil pressure predicted value, an engine temperature predicted value, a flight attitude predicted value and a flight speed predicted value output by 5 output nodes of the output layer respectively, W ₂₁、W₂₂、W₂₃、W₂₄、W₂₅ is the weight from the hidden layer to 5 output nodes on the output layer respectively, and b ₂₁、b₂₂、b₂₃、b₂₄、b₂₅ is the bias of 5 output nodes on the output layer respectively.

In this embodiment, the specific steps of performing model effect evaluation on the initial regression network model by using the mean square error algorithm are as follows:

s34, the mean square error algorithm is Wherein G is the number of flight data sets A _h in the model test set C, S _h is historical flight state data in the model test set C, and Y is flight state data predicted by an initial regression network model;

S35, presetting an evaluation value T, and taking the corresponding initial regression network model as a regression network model for predicting the SR20 type aircraft flight state data when the MSE is less than or equal to T.

In this embodiment, the state error calculation strategy includes: acquiring the current flight altitude of the SR20 type aircraft, real-time weather data and real-time flight state data at the current flight altitude, inputting the current flight altitude and the real-time weather data at the current flight altitude into a trained regression network model for predicting the SR20 type aircraft flight state data to obtain predicted flight state data Y＝[Y₁,Y₂,Y₃,Y₄,Y₅],Y₁、Y₂、Y₃、Y₄、Y₅ which are respectively engine parameter predicted values, oil pressure predicted values, engine temperature predicted values, flight attitude predicted values and flight speed predicted values, wherein the acquired real-time engine parameters, real-time oil pressure, real-time engine temperature, real-time flight attitude and real-time flight speed are respectively K ₁、K₂、K₃、K₄、K₅, respectively calculating an engine parameter state error D1, an oil pressure state error D2, an engine temperature state error D3, a flight attitude state error D4 and a flight speed state error D5, respectively presetting an engine parameter state error threshold U1, an oil pressure state error threshold U2, an engine temperature state error threshold U3, a flight attitude state error threshold U4 and a flight speed state error threshold U5, and a state error calculation formula beingAnd when Di is larger than Ui, carrying out abnormal state fault early warning on the ith data in the real-time flight state data corresponding to the ith state error, and feeding back abnormal component data corresponding to the abnormal state to the pilot.

Example 2

The embodiment provides an SR20 type aircraft fault detection system based on a visualization technology, and the specific scheme is that, as shown in fig. 2, the SR20 type aircraft fault detection system based on the visualization technology is implemented based on the SR20 type aircraft fault detection method based on the visualization technology described in embodiment 1, and the system comprises the following modules:

the data acquisition module is used for acquiring the multiple flying heights of the SR20 type aircraft, and acquiring historical flight state data of the SR20 type aircraft under different flying heights and historical weather data under different flying heights based on the multiple flying heights; acquiring real-time flight state data of an SR20 type airplane at the current flight altitude and real-time weather data of the SR20 type airplane at the current flight altitude;

The regression network model building module is used for training a regression network model for predicting the SR20 type aircraft flight state data based on the flight altitude, the historical flight state data and the historical weather data;

The model prediction module is used for inputting real-time weather data under the current flight altitude into the regression network model based on the regression network model to obtain SR20 type aircraft flight state data predicted under the current flight altitude;

The state error calculation module is used for substituting the SR20 type airplane flight state data predicted based on the current flight altitude and the real-time flight state data of the SR20 type airplane at the current flight altitude into a state error calculation strategy to calculate the state error in the SR20 type airplane real-time flight state data;

The fault early warning module is used for arranging the state errors in a descending order, presetting a state error threshold value, carrying out fault early warning on an abnormal state corresponding to the state error larger than the state error threshold value in the real-time flight state data, and feeding back abnormal component data corresponding to the abnormal state to a pilot;

and the control module is used for controlling the operation of each module.

The foregoing has shown and described the basic principles and main features of the present invention and the advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (8)

1. The SR20 type aircraft fault detection method based on the visualization technology is characterized by comprising the following steps of: the method comprises the following steps:

S1, acquiring multiple flying heights of an SR20 type airplane, and acquiring historical flight state data of the SR20 type airplane at different flying heights and historical weather data at different flying heights based on the multiple flying heights;

S2, acquiring real-time flight state data of the SR20 type aircraft at the current flight altitude and real-time weather data of the SR20 type aircraft at the current flight altitude;

S3, training a regression network model for predicting the SR20 type aircraft flight state data based on the flight altitude, the historical flight state data and the historical weather data;

s4, inputting real-time weather data under the current flight altitude into the regression network model based on the regression network model to obtain SR20 type aircraft flight state data predicted under the current flight altitude;

S5, substituting the SR20 type airplane flight state data predicted based on the current flight altitude and the real-time flight state data of the SR20 type airplane under the current flight altitude into a state error calculation strategy, and calculating the state error in the SR20 type airplane real-time flight state data;

s6, arranging the state errors in a descending order, presetting a state error threshold value, carrying out fault early warning on an abnormal state corresponding to the state error larger than the state error threshold value in the real-time flight state data, and feeding back abnormal component data corresponding to the abnormal state to a pilot.

2. The method for detecting the fault of the SR20 type aircraft based on the visualization technology according to claim 1, wherein the method comprises the following steps: the flight state data comprise engine parameters, oil pressure, engine temperature, flight attitude and flight speed; the weather data includes wind speed, wind direction, air humidity, visibility, temperature.

3. The method for detecting the fault of the SR20 type aircraft based on the visualization technology according to claim 2, wherein the method comprises the following steps: the specific content of S1 further includes:

And establishing a flight data set A _h＝{h,W_h,S_h when the flight height is h, wherein W _h is historical weather data when the flight height is h, S _h is historical flight state data when the flight height is h, respectively cleaning the flight data sets at different flight heights, checking whether missing values exist in the historical weather data and the historical flight state data, deleting the flight data sets at the corresponding flight heights containing the missing values if the missing values exist, and taking the flight data sets at all flight heights after the data cleaning as a model data set.

4. A method for detecting the fault of an SR20 type aircraft based on the visualization technique according to claim 3, characterized in that: the S3 further comprises the following concrete contents: dividing a model data set into a model training set and a model testing set, and constructing a regression network model; taking the flying height in the flying data set under all flying heights in the model training set, the historical wind speed, the historical wind direction, the historical air humidity, the historical visibility and the historical temperature in the historical weather data as inputs of a regression network model, taking the historical engine parameters, the historical oil pressure, the historical engine temperature, the historical flying posture and the historical flying speed in the historical flying state data in the flying data set under all flying heights in the model training set as outputs of the regression network model, training the model to obtain an initial regression network model, and evaluating the model effect of the initial regression network model by utilizing a mean square error algorithm and presetting an evaluation value; and screening a corresponding initial regression network model with the mean square error smaller than or equal to a preset evaluation value as a regression network model for predicting the SR20 type aircraft flight state data.

5. The method for detecting the fault of the SR20 type airplane based on the visualization technology, which is characterized by comprising the following steps of: the specific steps of obtaining the initial regression network model include:

S31, constructing a regression network model comprising an input layer, a hidden layer and an output layer, wherein the input layer comprises 6 input nodes, and the fly height, the wind speed, the wind direction, the air humidity, the visibility and the temperature are respectively input through the 6 input nodes; the output layer comprises 5 output nodes, and each output node respectively corresponds to and outputs an engine parameter predicted value, an oil pressure predicted value, an engine temperature predicted value, a flight attitude predicted value and a flight speed predicted value;

S32, the hidden layer output formula is h=σ (W ₁×X+b₁), where H is the hidden layer output, W ₁ is the hidden layer weight, b ₁ is the bias, σ is the activation function, where x= [ X ₁,X₂,X₃,X₄,X₅,X₆ ] is the input feature, X ₁ is the input flying height, X ₂ is the input wind speed, X ₃ is the input wind direction, X ₄ is the input air humidity, X ₅ is the input visibility, and X ₆ is the input temperature;

S33, the predicted value output by the output layer is predicted flight state data Y＝[W₂₁,W₂₂,W₂₃,W₂₄,W₂₅]×H+[b₂₁,b₂₂,b₂₃,b₂₄,b₂₅],, wherein Y＝[Y₁,Y₂,Y₃,Y₄,Y₅],Y₁、Y₂、Y₃、Y₄、Y₅ is an engine parameter predicted value, an oil pressure predicted value, an engine temperature predicted value, a flight attitude predicted value and a flight speed predicted value output by 5 output nodes of the output layer respectively, W ₂₁、W₂₂、W₂₃、W₂₄、W₂₅ is the weight from the hidden layer to 5 output nodes on the output layer respectively, and b ₂₁、b₂₂、b₂₃、b₂₄、b₂₅ is the bias of 5 output nodes on the output layer respectively.

6. The method for detecting the fault of the SR20 type airplane based on the visualization technology, which is characterized by comprising the following steps of: the specific steps of evaluating the model effect of the initial regression network model by using the mean square error algorithm are as follows:

s34, the mean square error algorithm is Wherein G is the number of flight data sets A _h in the model test set C, S _h is historical flight state data in the model test set C, and Y is flight state data predicted by an initial regression network model;

S35, presetting an evaluation value T, and taking the corresponding initial regression network model as a regression network model for predicting the SR20 type aircraft flight state data when the MSE is less than or equal to T.

7. The method for detecting the fault of the SR20 type airplane based on the visualization technology according to claim 6, wherein the method comprises the following steps: the state error calculation strategy comprises the following steps: acquiring the current flight altitude of the SR20 type aircraft, real-time weather data and real-time flight state data at the current flight altitude, inputting the current flight altitude and the real-time weather data at the current flight altitude into a trained regression network model for predicting the SR20 type aircraft flight state data to obtain predicted flight state data Y＝[Y₁,Y₂,Y₃,Y₄,Y₅],Y₁、Y₂、Y₃、Y₄、Y₅ which are respectively engine parameter predicted values, oil pressure predicted values, engine temperature predicted values, flight attitude predicted values and flight speed predicted values, wherein the acquired real-time engine parameters, real-time oil pressure, real-time engine temperature, real-time flight attitude and real-time flight speed are respectively K ₁、K₂、K₃、K₄、K₅, respectively calculating an engine parameter state error D1, an oil pressure state error D2, an engine temperature state error D3, a flight attitude state error D4 and a flight speed state error D5, respectively presetting an engine parameter state error threshold U1, an oil pressure state error threshold U2, an engine temperature state error threshold U3, a flight attitude state error threshold U4 and a flight speed state error threshold U5, and a state error calculation formula beingAnd when Di is larger than Ui, carrying out abnormal state fault early warning on the ith data in the real-time flight state data corresponding to the ith state error, and feeding back abnormal component data corresponding to the abnormal state to the pilot.

8. A visualization technology-based SR20 type aircraft fault detection system implemented based on the visualization technology-based SR20 type aircraft fault detection method as claimed in any one of claims 1 to 7, characterized in that: the system comprises the following modules:

the data acquisition module is used for acquiring the multiple flying heights of the SR20 type aircraft, and acquiring historical flight state data of the SR20 type aircraft under different flying heights and historical weather data under different flying heights based on the multiple flying heights; acquiring real-time flight state data of an SR20 type airplane at the current flight altitude and real-time weather data of the SR20 type airplane at the current flight altitude;

The regression network model building module is used for training a regression network model for predicting the SR20 type aircraft flight state data based on the flight altitude, the historical flight state data and the historical weather data;

The model prediction module is used for inputting real-time weather data under the current flight altitude into the regression network model based on the regression network model to obtain SR20 type aircraft flight state data predicted under the current flight altitude;

The state error calculation module is used for substituting the SR20 type airplane flight state data predicted based on the current flight altitude and the real-time flight state data of the SR20 type airplane at the current flight altitude into a state error calculation strategy to calculate the state error in the SR20 type airplane real-time flight state data;

The fault early warning module is used for arranging the state errors in a descending order, presetting a state error threshold value, carrying out fault early warning on an abnormal state corresponding to the state error larger than the state error threshold value in the real-time flight state data, and feeding back abnormal component data corresponding to the abnormal state to a pilot;

and the control module is used for controlling the operation of each module.