Emergency landing method based on eVTOL aircraft battery voltage abnormity

Abstract

The invention discloses an emergency landing method based on eVTOL aircraft battery voltage abnormity, and belongs to the technical field of eVTOL aircraft system detection. In order to guarantee the flight safety of the eVTOL aircraft, the method comprises the following steps: collecting the battery voltage of the eVTOL aircraft and the data of influence factors of the battery voltage of the eVTOL aircraft at the same moment, and constructing an aircraft battery voltage data prediction data set; constructing an eVTOL aircraft battery voltage data prediction model based on the improved LSTM neural network; performing voltage abnormal condition identification on the predicted voltage, and extracting voltage data of the voltage abnormal condition; constructing a voltage abnormal condition classification model, and performing voltage abnormal condition classification on the voltage data of the voltage abnormal condition; and based on different types of voltage abnormal conditions, an eVTOL aircraft emergency landing point selection method is constructed. According to the invention, the flight safety and the emergency response capability are greatly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of eVTOL aircraft system detection, and in particular relates to an emergency landing method based on abnormal battery voltage of an eVTOL aircraft. Background Art

[0002] Battery voltage stability is critical for eVTOL aircraft, as it directly affects flight safety, performance efficiency, system life, and energy management. Stable voltage ensures that the aircraft's motors and flight control systems can operate at their best, thereby improving flight smoothness and consistency of power output. In addition, voltage stability also helps extend the life of batteries and electronic components, maintaining device performance by reducing wear and tear caused by fluctuations. Effective battery management and energy usage prediction also rely on voltage stability, which is essential to ensure that the aircraft can complete its scheduled flight mission and return safely.

[0003] However, abnormal voltage fluctuations can lead to a series of serious consequences, including power system failure, control system failure, and even battery damage. These problems may lead to unstable aircraft power output, delayed control response, and in severe cases, may even cause the aircraft to make an emergency landing or an accident, endangering personnel safety.

[0004] eVTOL aircraft lack a method for identifying voltage anomalies and face many problems and risks. First, voltage anomalies may cause failure or unstable performance of key systems of the aircraft, such as the propulsion system, navigation system, and communication system, which directly threatens flight safety and increases the risk of accidents. For example, low voltage or large fluctuations may cause unstable power output of the motor, affect the lift and driving force of the aircraft, and may even cause the aircraft to lose control. Secondly, if voltage anomalies are not identified and handled in a timely manner, it may cause permanent damage to the electronic equipment of the aircraft, increase repair costs and difficulty of maintenance. Long-term voltage instability may also affect the health and service life of the battery, thereby affecting the economic benefits and operational efficiency of the entire aircraft. Summary of the invention

[0005] The problem to be solved by the present invention is to ensure the flight safety of the eVTOL aircraft, and propose an emergency landing method based on the abnormal battery voltage of the eVTOL aircraft.

[0006] To achieve the above object, the present invention is implemented through the following technical solutions: An emergency landing method based on abnormal battery voltage of an eVTOL aircraft comprises the following steps: S1. Collect data on the battery voltage of the eVTOL aircraft and the factors affecting the battery voltage of the eVTOL aircraft at the same time, and construct a data set for predicting the battery voltage of the aircraft, including a training set, a validation set, and a test set; S2. Construct an eVTOL aircraft battery voltage data prediction model based on an improved LSTM neural network; S3. Train the eVTOL aircraft battery voltage data prediction model using the training set obtained in step S1 to obtain a trained eVTOL aircraft battery voltage data prediction model; S4. Real-time data collection of the battery voltage of the eVTOL aircraft and the factors affecting the battery voltage of the eVTOL aircraft during flight, using the trained eVTOL aircraft battery voltage data prediction model obtained in step S3 to predict the battery voltage of the eVTOL aircraft, identifying voltage anomalies based on the predicted voltage, and extracting voltage data of the voltage anomalies; S5. Constructing a voltage abnormality classification model, classifying the voltage abnormality of the voltage data obtained in step S4; S6. Based on the different types of voltage anomalies obtained in step S5, a method for selecting an emergency landing point for an eVTOL aircraft is constructed.

[0007] Furthermore, the specific implementation method of step S1 includes the following steps: S1.1. Collect data on the battery voltage of the eVTOL aircraft and factors affecting the battery voltage of the eVTOL aircraft at the same time, wherein the data on factors affecting the battery voltage of the eVTOL aircraft include ambient temperature, number of charge and discharge cycles, flight speed, flight resistance, initial battery charge, flight altitude, and weight of the drone; Using infrared thermal imaging camera to capture the drone in flight The corresponding ambient temperatures at the time are ; By reading the recorded values ​​in the battery management system, the drone is able to The number of charge and discharge cycles corresponding to the time is ; By using the aircraft-mounted navigation module to collect data during the flight of the drone The flight speeds corresponding to the time are ; Measuring the pressure during the flight of the drone The flight resistance corresponding to the time is ; The coulomb meter is used to collect the data during the flight of the drone. The initial battery charge corresponding to the time is ; The navigation module on the aircraft collects the data during the flight of the drone. The flight altitudes corresponding to the time are ; The load sensor collects the data during the flight of the drone. The weight of the drone corresponding to the time is ; The voltage sensor is used to collect the voltage during the flight of the drone. The battery voltages corresponding to the time are ; S1.2. The result obtained in step S1.1 t 1~ tn The data of the moment is divided into multiple continuous time series, the length of each time series is J, and the number of divided time series is K; that is, the time corresponding to the first time series is: t1~tJ; the time corresponding to the second time series is: t2~t(J+1); the time corresponding to the third time series is: t3~t(J+2); ...; the time corresponding to the Kth time series is: tK~t(J+K-1); Divide all K time series data into training set, validation set, and test set in a ratio of 8:1:1.

[0008] Furthermore, the specific implementation method of step S2 includes the following steps: S2.1. Set the activation value calculation expression of the improved forget gate in the improved LSTM neural network as: ; in, is the activation value of the forget gate, is the sigmoid activation function, is layer normalization, is Dropout regularization, is the weight of the forget gate, is the final output of the previous time step, is the input of the current time step, For the connection operation, Forgotten Gate The weight of is the cell state at the previous time step, is the bias of the forget gate; S2.2. In the improved LSTM neural network, the activation value calculation method of the improved input gate is as follows: ; in, is the activation value of the input gate, is the weight of the input gate, For the input gate The weight of is the bias of the input gate; S2.3. In the improved LSTM neural network, the improved cell candidate value calculation method is as follows: ; in, is the cell candidate value, is the hyperbolic tangent activation function, is the weight of the cell candidate value, For cell candidate values The weight of is the bias of the cell candidate value; S2.4. In the improved LSTM neural network, the calculation method for the improved cell state update is as follows: ; in, is the attention weight, is the softmax activation function, is the coefficient of attention weight, is the attention weight The weight of is the bias of the attention weight; ; in, is the cell state; is the dot product operation; S2.5. In the improved LSTM neural network, the improved output gate and final output calculation method are as follows: ; in, is the activation value of the output gate, is the weight of the output gate, is the weight of the output gate, is the bias of the output gate; ; in, is the final output value; S2.6. Update the improved LSTM model calculation formula to the algorithm corresponding to the LSTM model and deploy it to the computer to complete the construction of the eVTOL aircraft battery voltage data prediction model.

[0009] Furthermore, the specific implementation method of step S3 includes the following steps: S3.1. The training set is input into the eVTOL aircraft battery voltage data prediction model for training. During the training process, the loss function uses the mean square error MSE, and the calculation method is as follows: ; in, is the sample size, is the actual value of the battery voltage, is the predicted value of battery voltage; S3.2. Error correction is achieved through back propagation and stochastic gradient descent (SGD) methods. When the loss function value tends to stabilize or reaches the preset training rounds, the training ends. The training of the eVTOL aircraft battery voltage data prediction model is completed, which is used to predict the battery voltage of the eVTOL aircraft during flight.

[0010] Furthermore, the method for identifying voltage abnormality for the predicted voltage in step S4 includes the following steps: Based on the prediction in step S3, Battery voltage value at the moment The battery voltage value actually monitored by the voltage sensor is ; Compare and ,when When is the allowable battery voltage error determined by the expert, indicating that the predicted The battery voltage value at the moment is different from the actual monitored The battery voltage value remains consistent at all times; when When The battery voltage value at the moment is different from the actual monitored The battery voltage values ​​at different times are quite different, and there is a voltage abnormality. The battery voltage value at time and M times before and after it.

[0011] Furthermore, the specific implementation method of step S5 includes the following steps: S5.1. Set voltage abnormal conditions including voltage sag, continuous voltage drop, abnormal voltage fluctuation, and instantaneous voltage spike; S5.2. Obtain voltage sag, voltage continuous drop, voltage abnormal fluctuation, and instantaneous voltage spike data from the aviation battery performance database of the industry database, and use the data to draw a corresponding discounted graph of voltage change over time; S5.3. Use the labeling tool LabelImg to label the voltage anomaly in the discounted graph of voltage change over time obtained in step S5.2, where the labeling information includes the coordinates of the labeling box and the category of the anomaly; divide the labeled data into a training set, a validation set, and a test set according to the ratio of 70%, 15%, and 15%; S5.4. Deploy the YOLOv9 deep learning model in the computer, set the learning rate and batch size hyperparameters of the model based on the training set, validation set and test set divided in step S5.3, use the optimizer and loss function to train the YOLOv9 deep learning model, and when the model loss function value tends to be stable or reaches the preset training rounds, the training ends and the voltage abnormality classification model is obtained; S5.5. Draw a line graph of the battery voltage value changing over time for the voltage data of the voltage abnormality obtained in step S4; then input it into the voltage abnormality classification model to classify the voltage abnormality.

[0012] Furthermore, the specific implementation method of step S6 includes the following steps: S6.1. Construct an emergency landing point selection method for eVTOL aircraft; S6.1.1. Use the eVTOL aircraft’s onboard navigation system to obtain the aircraft’s current location and current speed , using a coulomb meter to measure the remaining battery capacity , using current sensor to measure current , using a voltage sensor to measure voltage , on this basis, calculate the current power consumption , the calculation formula is: ; Then calculate the remaining flight distance of the eVTOL aircraft , the calculation formula is: ; S6.1.2. Emergency landing point selection: Use geographic information systems to identify potential landing sites within the eVTOL aircraft's field of view and select nm potential landing sites; considerations for potential landing sites include terrain flatness PT , obstacle density ZM , population density RM Surface type DB , the consideration data of landing point are obtained from the geographic information system platform as follows: A scoring standard is established for each potential landing point, and the calculation expression is: ; in, For the The score corresponding to the potential landing point, are the weight coefficients of terrain flatness, obstacle density, population density, and surface type, Respectively The scores of terrain flatness, obstacle density, population density, and surface type corresponding to each potential landing point are obtained by direct scoring by staff; In order of rating from highest to lowest, Sort the potential landing points, and calculate the potential landing points in sequence according to the sorting The current position of the aircraft The distance between , the calculation formula is: ; Compare and The relationship between: When , it indicates that the currently calculated potential landing point is within the remaining flight distance of the eVTOL aircraft, that is, the currently calculated potential landing point is selected as the emergency landing point; when This indicates that the currently calculated potential landing point is not within the remaining flight distance of the eVTOL aircraft. You need to continue sorting the scores from large to small and calculate the distance between the next potential landing point and the current position of the aircraft until the score is satisfied. , the currently calculated potential landing point is selected as the emergency landing point, and then the calculation is stopped; S6.2. If the voltage abnormality obtained in step S5 is classified as voltage drop, the eVTOL aircraft emergency landing point selection method is used to perform emergency landing. If it is not, the battery status is recorded and the battery is repaired.

[0013] Beneficial effects of the present invention: The present invention discloses an emergency landing method based on abnormal battery voltage of an eVTOL aircraft. During the flight of the VTOL aircraft, the voltage data of the battery is monitored in real time to determine the abnormal voltage situation. The abnormal voltage situation identified is then classified using an abnormal voltage classification model. Finally, in view of the situation where the battery is severely damaged, an emergency landing point selection method for the eVTOL aircraft is proposed. Ultimately, comprehensive monitoring of the battery voltage during the flight of the eVTOL aircraft can be achieved to ensure the flight safety of the aircraft. The present invention can comprehensively and accurately realize the identification of abnormal voltage values, thereby discovering and preventing potential faults of the power system in advance, ensuring the normal operation of the equipment and the stability of the system, and reducing the risk of unexpected power outages.

[0014] The emergency landing method based on abnormal battery voltage of an eVTOL aircraft described in the present invention can quickly determine the emergency landing point based on the surrounding environment information after a serious abnormality occurs in the aircraft voltage, thereby greatly improving flight safety and emergency response capabilities and reducing the risk of accidents caused by voltage problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of an emergency landing method based on abnormal battery voltage of an eVTOL aircraft according to the present invention; DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the specific embodiments described are only part of the embodiments of the present invention, rather than all of the specific embodiments. The components of the specific embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations, and the present invention can also have other embodiments.

[0017] Therefore, the following detailed description of the specific embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected specific embodiments of the present invention. Based on the specific embodiments of the present invention, all other specific embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0018] In order to further understand the content, features and effects of the present invention, the following specific implementation methods are given as examples, and the attached Figure 1 The detailed instructions are as follows:

[0019] Embodiment 1: An emergency landing method based on abnormal battery voltage of an eVTOL aircraft comprises the following steps: S1. Collect data on the battery voltage of the eVTOL aircraft and the factors affecting the battery voltage of the eVTOL aircraft at the same time, and construct a data set for predicting the battery voltage of the aircraft, including a training set, a validation set, and a test set; Furthermore, the specific implementation method of step S1 includes the following steps: S1.1. Collect data on the battery voltage of the eVTOL aircraft and factors affecting the battery voltage of the eVTOL aircraft at the same time, wherein the data on factors affecting the battery voltage of the eVTOL aircraft include ambient temperature, number of charge and discharge cycles, flight speed, flight resistance, initial battery charge, flight altitude, and weight of the drone; Using infrared thermal imaging camera to capture the drone in flight The corresponding ambient temperatures at the time are ; Furthermore, the ambient temperature refers to the ambient temperature of the drone's operating environment. Low temperatures will increase the battery's internal resistance, reduce discharge efficiency, and cause voltage drop; high temperatures may cause the battery to overheat, affecting battery performance and life.

[0020] By reading the recorded values ​​in the battery management system, the drone is able to The number of charge and discharge cycles corresponding to the time is ; Furthermore, the number of charge and discharge cycles refers to the number of times a battery completes a full cycle of charge and discharge. Each time a battery is used from full to needing to be recharged and then fully charged, it is counted as one cycle. As the number of charge and discharge cycles increases, the battery capacity gradually decreases, resulting in a faster voltage drop under the same load.

[0021] By using the aircraft-mounted navigation module to collect data during the flight of the drone The flight speeds corresponding to the time are ; Furthermore, flight speed refers to the rate at which the drone moves in the air. Flying at high speeds increases wind resistance, requiring more power to overcome resistance, resulting in greater current consumption and faster voltage drop.

[0022] Measuring the pressure during the flight of the drone The flight resistance corresponding to the time is ; Furthermore, flight resistance refers to the force that hinders the movement of the drone in the air. It mainly includes air resistance. Flying against the wind consumes more energy than flying with the wind, and the acceleration voltage decreases.

[0023] The coulomb meter is used to collect the data during the flight of the drone. The initial battery charge corresponding to the time is ; Furthermore, the initial charge of the battery refers to the battery's charge status before the drone takes off. It is usually expressed as a percentage, with 100% indicating that the battery is fully charged. Insufficient initial charge will result in a lower voltage at takeoff, and the voltage will drop faster during flight. Overcharging may shorten the battery life and affect long-term voltage stability.

[0024] The navigation module on the aircraft collects the data during the flight of the drone. The flight altitudes corresponding to the time are ; Furthermore, flight altitude refers to the vertical distance of the drone relative to the ground or sea level. When flying at high altitudes, the air density decreases, requiring more power to maintain flight, accelerating the voltage drop. Frequent altitude changes (ascending and descending) increase energy consumption and affect voltage stability.

[0025] The load sensor collects the data during the flight of the drone. The weight of the drone corresponding to the time is ; Furthermore, drone weight refers to the total weight of the drone, including the sum of the weights of all components such as the drone body, battery, and payload. An increase in the total weight of the drone will directly increase the motor load, resulting in greater current consumption and faster voltage drop. Uneven weight distribution may cause some motors to be overloaded, resulting in uneven voltage consumption.

[0026] The voltage sensor is used to collect the voltage during the flight of the drone. The battery voltages corresponding to the time are ; S1.2. The result obtained in step S1.1 t 1~ tn The data of the moment is divided into multiple continuous time series, the length of each time series is J, and the number of divided time series is K; that is, the time corresponding to the first time series is: t1~tJ; the time corresponding to the second time series is: t2~t(J+1); the time corresponding to the third time series is: t3~t(J+2); ...; the time corresponding to the Kth time series is: tK~t(J+K-1); Divide all K time series data into training set, validation set, and test set in a ratio of 8:1:1.

[0027] S2. Construct an eVTOL aircraft battery voltage data prediction model based on an improved LSTM neural network; The voltage sensor is used to measure and monitor the voltage of the battery terminal in real time during the flight of the eVTOL aircraft. By building a battery voltage data prediction model, the battery voltage data during the flight of the eVTOL aircraft can be predicted, and then compared with the monitoring data to identify abnormal voltage data.

[0028] Construction of a battery voltage prediction model: There are two problems in the process of building a prediction model: (1) It is difficult to capture the long-term dependencies between data in traditional models, that is, it is difficult for the model to remember and use the battery voltage values ​​at earlier times to predict the battery voltage values ​​at future times. (2) When predicting the battery voltage value, not all information at earlier times is equally important, and it is necessary to distinguish the importance of the information at existing times.

[0029] To address the above problems, the LSTM (Long Short-Term Memory) model was proposed. This model introduces a gating mechanism, including an input gate, a forget gate, and an output gate. By controlling the long-term preservation and forgetting of information, it effectively solves the problem of long-term data retention. At the same time, the gating mechanism of LSTM allows the model to learn the importance of information at different time nodes. However, there are still the following problems in LSTM: (1) When training LSTM, the performance of the model may be affected due to the large fluctuations in the data in the time series, especially when the data is unevenly distributed. (2) Traditional LSTM models may not be accurate enough when processing data with complex time dependencies, such as changes in battery voltage values, because this method usually only relies on short-term input and hidden states. (3) When processing long sequence data, traditional LSTM may ignore information that is particularly important for prediction. This method is usually used to process information of the entire sequence. (4) When training complex LSTM networks, the model will overfit the training data, resulting in performance degradation in practical applications.

[0030] The above problems will make it difficult to accurately predict the battery voltage value based on monitoring data during the flight of the eVTOL aircraft.

[0031] Therefore, in the present invention, the LSTM model is improved. The details are as follows: (1) Introducing layer normalization technology. Layer normalization can stabilize the learning process of LSTM and improve the adaptability of the model under different data conditions, which is very important for drones flying in different environments and load conditions. This technology can help the model better adapt to environmental changes, such as the impact of ambient temperature and load changes on battery voltage values.

[0032] (2) Introducing peak connection technology. Peak connection allows the model to directly use long-term memory to adjust the opening and closing of the gate, which is very useful for predicting the voltage value of a battery under continuous operation.

[0033] (3) Introducing the attention mechanism. The attention mechanism can help the model identify the flight phase that has the most critical impact on battery voltage changes.

[0034] (4) Introducing Dropout regularization technology. In the prediction model of the battery voltage value of the drone, the use of Dropout can increase the generalization ability of the model and reduce the prediction error under new or unseen flight conditions. This helps to maintain the stability and accuracy of the prediction when the drone faces different environmental conditions or flight modes.

[0035] The above four technologies are used to improve the forget gate, input gate, cell candidate value, cell state update value, output gate, and final output in the traditional LSTM, as follows: The cell state is the core part of the network memory, which transmits information through a series of time steps. The cell state can be regarded as a flowing "information highway" that carries all the information that may be important for future decisions. In LSTM, this cell state can be transmitted in an almost unchanged form between different time steps, and only slightly adjusted through gate control.

[0036] Furthermore, the specific implementation method of step S2 includes the following steps: S2.1. Set the activation value calculation expression of the improved forget gate in the improved LSTM neural network as: ; in, is the activation value of the forget gate, is the sigmoid activation function, is layer normalization, is Dropout regularization, is the weight of the forget gate, is the final output of the previous time step, is the input of the current time step, For the connection operation, Forgotten Gate The weight of is the cell state at the previous time step, is the bias of the forget gate; Furthermore, the forget gate determines the information to be discarded from the cell state. It calculates a value between 0 and 1 through a sigmoid activation function, which is multiplied element by element with the cell state to achieve "forgetting" of the state. For the battery voltage prediction during the flight of the eVTOL aircraft, if the battery voltage does not change much under certain conditions, the forget gate can help the model ignore these unimportant changes.

[0037] S2.2. In the improved LSTM neural network, the activation value calculation method of the improved input gate is as follows: ; in, is the activation value of the input gate, is the weight of the input gate, For the input gate The weight of is the bias of the input gate; Furthermore, the input gate controls the importance of new information and determines the new information that will be added to the cell state. The cell candidate value is a transformed form of the new information that may be added to the cell state. For the battery voltage value prediction during the flight of the eVTOL aircraft, the input gate controls the importance of the new voltage input information, and the cell candidate value updates the voltage value state based on this information.

[0038] S2.3. In the improved LSTM neural network, the improved cell candidate value calculation method is as follows: ; in, is the cell candidate value, is the hyperbolic tangent activation function, is the weight of the cell candidate value, For cell candidate values The weight of is the bias of the cell candidate value; Furthermore, the cell state is where LSTM stores past information, and it is updated at each time step to maintain long-term dependent information. When predicting the battery voltage value during the flight of the eVTOL aircraft, the update of the cell state allows the model to retain and forget information, ensuring that only useful information is retained, thereby improving the accuracy of the prediction.

[0039] S2.4. In the improved LSTM neural network, the calculation method for the improved cell state update is as follows: ; in, is the attention weight, is the softmax activation function, is the coefficient of attention weight, is the attention weight The weight of is the bias of the attention weight; ; in, is the cell state; is the dot product operation; S2.5. In the improved LSTM neural network, the improved output gate and final output calculation method are as follows: ; in, is the activation value of the output gate, is the weight of the output gate, is the weight of the output gate, is the bias of the output gate; ; in, is the final output value; Furthermore, the output gate controls the amount of information output from the LSTM unit. The final output is based on the current cell state and the output of the output gate. It is the output passed to the next time step and the final output. For the prediction of the battery voltage value during the flight of the eVTOL aircraft, the output gate and the final output help determine the voltage value of each step and provide data support for battery management.

[0040] S2.6. Update the improved LSTM model calculation formula to the algorithm corresponding to the LSTM model and deploy it to the computer to complete the construction of the eVTOL aircraft battery voltage data prediction model.

[0041] S3. Train the eVTOL aircraft battery voltage data prediction model using the training set obtained in step S1 to obtain a trained eVTOL aircraft battery voltage data prediction model; Furthermore, the specific implementation method of step S3 includes the following steps: S3.1. The training set is input into the eVTOL aircraft battery voltage data prediction model for training. During the training process, the loss function uses the mean square error MSE, and the calculation method is as follows: ; in, is the sample size, is the actual value of the battery voltage, is the predicted value of battery voltage; S3.2. Error correction is achieved through back propagation and stochastic gradient descent (SGD) methods. When the loss function value tends to stabilize or reaches the preset training rounds, the training ends. The training of the eVTOL aircraft battery voltage data prediction model is completed, which is used to predict the battery voltage of the eVTOL aircraft during flight.

[0042] S4. Real-time data collection of the battery voltage of the eVTOL aircraft and the factors affecting the battery voltage of the eVTOL aircraft during flight, using the trained eVTOL aircraft battery voltage data prediction model obtained in step S3 to predict the battery voltage of the eVTOL aircraft, identifying voltage anomalies based on the predicted voltage, and extracting voltage data of the voltage anomalies; Furthermore, the method for identifying voltage abnormality for the predicted voltage in step S4 includes the following steps: Based on the prediction in step S3, Battery voltage value at the moment The battery voltage value actually monitored by the voltage sensor is ; Compare and ,when When is the allowable battery voltage error determined by the expert, indicating that the predicted The battery voltage value at the moment is different from the actual monitored The battery voltage value remains consistent at all times; when When The battery voltage value at the moment is different from the actual monitored The battery voltage values ​​at different times are quite different, and there is a voltage abnormality. The battery voltage value at time and M times before and after it.

[0043] S5. Constructing a voltage abnormality classification model, classifying the voltage abnormality of the voltage data obtained in step S4; Furthermore, the specific implementation method of step S5 includes the following steps: S5.1. Set voltage abnormal conditions including voltage sag, continuous voltage drop, abnormal voltage fluctuation, and instantaneous voltage spike; Furthermore, abnormal voltage conditions include voltage sag, continuous voltage drop, abnormal voltage fluctuation, and instantaneous voltage spike. These types of abnormal voltage conditions have obvious abnormal characteristics and can be identified and classified through image recognition.

[0044] Voltage sag is a significant and sudden drop in battery voltage in a very short period of time. Voltage sag appears as an almost vertical line segment on the voltage-time graph. It is usually a sharp, almost instantaneous drop that may rebound quickly (forming a V-shape) or remain low for a period of time. Causes of voltage sag include: instantaneous high current load, internal battery short circuit, loose connection or poor contact.

[0045] Continuous voltage drop means that the battery voltage continues to drop at a faster rate than normal discharge. The continuous voltage drop appears as a decreasing curve with a large slope on the voltage-time graph; the decreasing speed is significantly faster than the normal discharge curve; it may be a linear decrease or a nonlinear accelerated decrease. The reasons for the continuous voltage drop include: battery capacity depletion, battery aging or damage, and continuous excessive load.

[0046] Abnormal voltage fluctuation refers to frequent and large fluctuations in the battery voltage within a short period of time. Abnormal voltage fluctuations appear as irregular sawtooth or wavy curves on the voltage-time graph; the fluctuation amplitude is significantly larger than the small fluctuation under normal working conditions; the fluctuation frequency may be high, but not necessarily regular. The causes of abnormal voltage fluctuations include: rapid changes in load and unstable electrochemical reactions inside the battery.

[0047] A transient voltage spike is a significant upward peak in the battery voltage in a very short time. A transient voltage spike appears as a sharp upward "spike" on the voltage-time graph; it is usually a very short spike, and then quickly falls back to normal levels; the peak value may be much higher than the normal operating voltage. The causes of transient voltage spikes include: sudden removal of load, electromagnetic interference, charging system failure, sensor or data acquisition system abnormality.

[0048] S5.2. Obtain voltage sag, voltage continuous drop, voltage abnormal fluctuation, and instantaneous voltage spike data from the aviation battery performance database of the industry database, and use the data to draw a corresponding discounted graph of voltage change over time; S5.3. Use the labeling tool LabelImg to label the voltage anomaly in the discounted graph of voltage change over time obtained in step S5.2, where the labeling information includes the coordinates of the labeling box and the category of the anomaly; divide the labeled data into a training set, a validation set, and a test set according to the ratio of 70%, 15%, and 15%; S5.4. Deploy the YOLOv9 deep learning model in the computer, set the learning rate and batch size hyperparameters of the model based on the training set, validation set and test set divided in step S5.3, use the optimizer and loss function to train the YOLOv9 deep learning model, and when the model loss function value tends to be stable or reaches the preset training rounds, the training ends and the voltage abnormality classification model is obtained; S5.5. Draw a line graph of the battery voltage value changing over time for the voltage data of the voltage abnormality obtained in step S4; then input it into the voltage abnormality classification model to classify the voltage abnormality.

[0049] S6. Based on the different types of voltage anomalies obtained in step S5, a method for selecting an emergency landing point for an eVTOL aircraft is constructed.

[0050] Furthermore, the specific implementation method of step S6 includes the following steps: S6.1. Construct an emergency landing point selection method for eVTOL aircraft; S6.1.1. Use the eVTOL aircraft’s onboard navigation system to obtain the aircraft’s current location and current speed , using a coulomb meter to measure the remaining battery capacity , using current sensor to measure current , using a voltage sensor to measure voltage , on this basis, calculate the current power consumption , the calculation formula is: ; Then calculate the remaining flight distance of the eVTOL aircraft , the calculation formula is: ; S6.1.2. Emergency landing point selection: Use geographic information systems to identify potential landing sites within the eVTOL aircraft's field of view and select nm potential landing sites; considerations for potential landing sites include terrain flatness PT , obstacle density ZM , population density RM Surface type DB , the consideration data of landing point are obtained from the geographic information system platform as follows: A scoring standard is established for each potential landing point, and the calculation expression is: ; in, For the The score corresponding to the potential landing point, are the weight coefficients of terrain flatness, obstacle density, population density, and surface type, Respectively The scores of terrain flatness, obstacle density, population density, and surface type corresponding to each potential landing point are obtained by direct scoring by staff; In order of rating from highest to lowest, Sort the potential landing points, and calculate the potential landing points in sequence according to the sorting The current position of the aircraft The distance between , the calculation formula is: ; Compare and The relationship between: When , it indicates that the currently calculated potential landing point is within the remaining flight distance of the eVTOL aircraft, that is, the currently calculated potential landing point is selected as the emergency landing point; when This indicates that the currently calculated potential landing point is not within the remaining flight distance of the eVTOL aircraft. You need to continue sorting the scores from large to small and calculate the distance between the next potential landing point and the current position of the aircraft until the score is satisfied. , the currently calculated potential landing point is selected as the emergency landing point, and then the calculation is stopped; S6.2. If the voltage abnormality obtained in step S5 is classified as voltage drop, the eVTOL aircraft emergency landing point selection method is used to perform emergency landing. If it is not, the battery status is recorded and the battery is repaired.

[0051] Furthermore, compared with continuous voltage drop, abnormal voltage fluctuation, and instantaneous voltage spike, voltage sag is considered to be the most serious and most likely to cause a crash. The danger of voltage sag mainly stems from its suddenness, comprehensive impact, and difficulty in recovery. It may cause the propulsion system and control system to fail at the same time, causing the aircraft to lose lift and control capabilities instantly. Once a voltage sag occurs, it becomes a vital safety measure to immediately find an emergency landing point. This is because time is tight and the system may fail quickly. Landing quickly can prevent the aircraft from completely losing control, minimize the risk of crashing, and effectively use the remaining power for a controlled landing. Dealing with voltage sag requires quick decision-making, selecting the best landing point, and executing emergency procedures. Therefore, when the above-mentioned voltage anomaly intelligent classification results are continuous voltage drop, abnormal voltage fluctuation, and instantaneous voltage spike, the battery needs to be repaired. During the battery inspection, if the battery damage is not serious, it can be repaired directly, for example, due to changes in ambient temperature, slight charge and discharge cycle effects, or short-term load fluctuations. If the battery is found to be seriously damaged, such as internal short circuit, electrolyte leakage, or battery expansion, an emergency landing point needs to be selected for emergency landing of the drone. When the above-mentioned voltage abnormality intelligent classification result is a voltage drop, an emergency landing point needs to be selected immediately.

[0052] The key points and intended protection points of the present invention are: (1) A method for identifying and classifying abnormal voltage conditions (2) A method for selecting an emergency landing point for an eVTOL aircraft.

[0053] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0054] Although the present application has been described above with reference to specific embodiments, various modifications may be made thereto and parts thereof may be replaced with equivalents without departing from the scope of the present application. In particular, as long as there is no structural conflict, the various features in the specific embodiments disclosed in the present application may be used in combination with each other in any manner, and the fact that these combinations are not exhaustively described in this specification is only for the sake of omitting space and saving resources. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. An emergency landing method based on abnormal battery voltage of an eVTOL aircraft, characterized in that: The steps include: S1. Collect data on the battery voltage of the eVTOL aircraft and the factors affecting the battery voltage of the eVTOL aircraft at the same time, and construct a data set for predicting the battery voltage of the aircraft, including a training set, a validation set, and a test set; S2. Construct an eVTOL aircraft battery voltage data prediction model based on an improved LSTM neural network; S3. Train the eVTOL aircraft battery voltage data prediction model using the training set obtained in step S1 to obtain a trained eVTOL aircraft battery voltage data prediction model; S4. Real-time data collection of the battery voltage of the eVTOL aircraft and the factors affecting the battery voltage of the eVTOL aircraft during flight, using the trained eVTOL aircraft battery voltage data prediction model obtained in step S3 to predict the battery voltage of the eVTOL aircraft, identifying voltage anomalies based on the predicted voltage, and extracting voltage data of the voltage anomalies; S5. Constructing a voltage abnormality classification model, classifying the voltage abnormality of the voltage data obtained in step S4; S6. Based on the different types of voltage anomalies obtained in step S5, a method for selecting an emergency landing point for an eVTOL aircraft is constructed.

2. The method for emergency landing based on abnormal battery voltage of an eVTOL aircraft according to claim 1, characterized in that: The specific implementation method of step S1 includes the following steps: S1.

1. Collect data on the battery voltage of the eVTOL aircraft and factors affecting the battery voltage of the eVTOL aircraft at the same time, wherein the data on factors affecting the battery voltage of the eVTOL aircraft include ambient temperature, number of charge and discharge cycles, flight speed, flight resistance, initial battery charge, flight altitude, and weight of the drone; Using infrared thermal imaging camera to capture the drone in flight The corresponding ambient temperatures at the time are ; By reading the recorded values ​​in the battery management system, the drone is able to The number of charge and discharge cycles corresponding to the time is ; By using the aircraft-mounted navigation module to collect data during the flight of the drone The flight speeds corresponding to the time are ; Measuring the pressure during the flight of the drone The flight resistance corresponding to the time is ; The coulomb meter is used to collect the data during the flight of the drone. The initial battery charge corresponding to the time is ; The navigation module on the aircraft collects the data during the flight of the drone. The flight altitudes corresponding to the time are ; The load sensor collects the data during the flight of the drone. The weight of the drone corresponding to the time is ; The voltage sensor is used to collect the voltage during the flight of the drone. The battery voltages corresponding to the time are ; S1.

2. The result obtained in step S1.1 t 1~ tn The data of the moment is divided into multiple continuous time series, the length of each time series is J, and the number of divided time series is K; that is, the time corresponding to the first time series is: t1~tJ; the time corresponding to the second time series is: t2~t(J+1); the time corresponding to the third time series is: t3~t(J+2); ...; the time corresponding to the Kth time series is: tK~t(J+K-1); Divide all K time series data into training set, validation set, and test set in a ratio of 8:1:

1.

3. The method for emergency landing based on abnormal battery voltage of an eVTOL aircraft according to claim 2, characterized in that: The specific implementation method of step S2 includes the following steps: S2.

1. Set the activation value calculation expression of the improved forget gate in the improved LSTM neural network as: ； in, is the activation value of the forget gate, is the sigmoid activation function, is layer normalization, is Dropout regularization, is the weight of the forget gate, is the final output of the previous time step, is the input of the current time step, For the connection operation, Forgotten Gate The weight of is the cell state at the previous time step, is the bias of the forget gate; S2.

2. In the improved LSTM neural network, the activation value calculation method of the improved input gate is as follows: ； in, is the activation value of the input gate, is the weight of the input gate, For the input gate The weight of is the bias of the input gate; S2.

3. In the improved LSTM neural network, the improved cell candidate value calculation method is as follows: ； in, is the cell candidate value, is the hyperbolic tangent activation function, is the weight of the cell candidate value, For cell candidate values The weight of is the bias of the cell candidate value; S2.

4. In the improved LSTM neural network, the calculation method for the improved cell state update is as follows: ； in, is the attention weight, is the softmax activation function, is the coefficient of attention weight, is the attention weight The weight of is the bias of the attention weight; ； in, is the cell state; is the dot product operation; S2.

5. In the improved LSTM neural network, the improved output gate and final output calculation method are as follows: ； in, is the activation value of the output gate, is the weight of the output gate, is the weight of the output gate, is the bias of the output gate; ； in, is the final output value; S2.

6. Update the improved LSTM model calculation formula to the algorithm corresponding to the LSTM model and deploy it to the computer to complete the construction of the eVTOL aircraft battery voltage data prediction model.

4. The method for emergency landing based on abnormal battery voltage of an eVTOL aircraft according to claim 3, characterized in that: The specific implementation method of step S3 includes the following steps: S3.

1. The training set is input into the eVTOL aircraft battery voltage data prediction model for training. During the training process, the loss function uses the mean square error MSE, and the calculation method is as follows: ； in, is the sample size, is the actual value of the battery voltage, is the predicted value of battery voltage; S3.

2. Error correction is achieved through back propagation and stochastic gradient descent (SGD) methods. When the loss function value tends to stabilize or reaches the preset training rounds, the training ends. The training of the eVTOL aircraft battery voltage data prediction model is completed, which is used to predict the battery voltage of the eVTOL aircraft during flight.

5. The method for emergency landing based on abnormal battery voltage of an eVTOL aircraft according to claim 4, characterized in that: The method for identifying abnormal voltage conditions on the predicted voltage in step S4 comprises the following steps: Based on the prediction in step S3, Battery voltage value at the moment The battery voltage value actually monitored by the voltage sensor is ; Compare and ,when When is the allowable battery voltage error determined by the expert, indicating that the predicted The battery voltage value at the moment is different from the actual monitored The battery voltage value remains consistent at all times; when When The battery voltage value at the moment is different from the actual monitored The battery voltage values ​​at different times are quite different, and there is a voltage abnormality. The battery voltage value at time and M times before and after it.

6. The method for emergency landing based on abnormal battery voltage of an eVTOL aircraft according to claim 5, characterized in that: The specific implementation method of step S5 includes the following steps: S5.

1. Set voltage abnormal conditions including voltage sag, continuous voltage drop, abnormal voltage fluctuation, and instantaneous voltage spike; S5.

2. Obtain voltage sag, voltage continuous drop, voltage abnormal fluctuation, and instantaneous voltage spike data from the aviation battery performance database of the industry database, and use the data to draw a corresponding discounted graph of voltage change over time; S5.

3. Use the labeling tool LabelImg to label the voltage anomaly in the discounted graph of voltage change over time obtained in step S5.2, where the labeling information includes the coordinates of the labeling box and the category of the anomaly; divide the labeled data into a training set, a validation set, and a test set according to the ratio of 70%, 15%, and 15%; S5.

4. Deploy the YOLOv9 deep learning model in the computer, set the learning rate and batch size hyperparameters of the model based on the training set, validation set and test set divided in step S5.3, use the optimizer and loss function to train the YOLOv9 deep learning model, and when the model loss function value tends to be stable or reaches the preset training rounds, the training ends and the voltage abnormality classification model is obtained; S5.

5. Draw a line graph of the battery voltage value changing over time for the voltage data of the voltage abnormality obtained in step S4; then input it into the voltage abnormality classification model to classify the voltage abnormality.

7. The method for emergency landing based on abnormal battery voltage of an eVTOL aircraft according to claim 6, characterized in that: The specific implementation method of step S6 includes the following steps: S6.

1. Construct an emergency landing point selection method for eVTOL aircraft; S6.1.

1. Use the eVTOL aircraft’s onboard navigation system to obtain the aircraft’s current location and current speed , using a coulomb meter to measure the remaining battery capacity , using current sensor to measure current , using a voltage sensor to measure voltage , on this basis, calculate the current power consumption , the calculation formula is: ； Then calculate the remaining flight distance of the eVTOL aircraft , the calculation formula is: ； S6.1.

2. Emergency landing point selection: Use geographic information systems to identify potential landing sites within the eVTOL aircraft's field of view and select nm potential landing sites; considerations for potential landing sites include terrain flatness PT , obstacle density ZM , population density RM Surface type DB , the consideration data of landing point are obtained from the geographic information system platform as follows: A scoring standard is established for each potential landing point, and the calculation expression is: ； in, For the The score corresponding to the potential landing point, are the weight coefficients of terrain flatness, obstacle density, population density, and surface type, Respectively The scores of terrain flatness, obstacle density, population density, and surface type corresponding to each potential landing point are obtained by direct scoring by staff; In order of rating from highest to lowest, Sort the potential landing points, and calculate the potential landing points in sequence according to the sorting The current position of the aircraft The distance between , the calculation formula is: ； Compare and The relationship between: When , it indicates that the currently calculated potential landing point is within the remaining flight distance of the eVTOL aircraft, that is, the currently calculated potential landing point is selected as the emergency landing point; when This indicates that the currently calculated potential landing point is not within the remaining flight distance of the eVTOL aircraft. You need to continue sorting the scores from large to small and calculate the distance between the next potential landing point and the current position of the aircraft until the score is satisfied. , the currently calculated potential landing point is selected as the emergency landing point, and then the calculation is stopped; S6.

2. If the voltage abnormality obtained in step S5 is classified as voltage drop, the eVTOL aircraft emergency landing point selection method is used to perform emergency landing. If it is not, the battery status is recorded and the battery is repaired.

Images