Avionics equipment and signal processing method based on high-precision inertial navigation

Abstract

The present invention provides an avionics device and a signal processing method based on high-precision inertial navigation, which relate to the field of aviation navigation control technology. The invention uses environmental complexity assessment to drive sensor selection, optimizes the acquisition quality of inertial data such as angular velocity and acceleration, and pixel coordinates, and optical data such as specific data in real time, reduces redundancy, utilizes spectral decomposition and adaptive filtering, dynamically cleans signals for environmental noise in different frequency bands, retains key features, uses spatiotemporal feature reconstruction to generate high-dimensional feature vectors, enhances the expressiveness of low-quality data, dynamically adjusts fusion weights based on signal-to-noise ratio and consistency index, combines predictive error correction to eliminate long-term drift, ensures processing delay is less than 10ms with hardware support, and maintains data stability through anti-interference shielding. The invention breaks through the dependence of traditional navigation on external signals, realizes long-term high-precision navigation, is particularly suitable for aviation covert missions and civil aviation flights in severe weather, and improves mission success rate and safety.

Description

Technical Field

[0001] The present invention relates to the technical field of aviation navigation control, and in particular to an avionics device based on high-precision inertial navigation and a signal processing method. Background Art

[0002] According to China Publication No. "CN118534472A," a method for autonomous navigation of an aircraft flying on the sea surface based on laser ranging horizon measurement is disclosed. The method includes the following steps: performing multi-beam ranging of the sea surface below the track using a laser ranging device installed on the aircraft to obtain a horizon information measurement value based on the ranging value; establishing an aircraft navigation error state model based on the aircraft inertial navigation mechanics equation as a navigation system state space representation required for Kalman filtering; establishing an aircraft navigation system measurement equation required for Kalman filtering based on the aircraft navigation error state model; performing extended Kalman filtering calculation based on the horizon information measurement value, the aircraft navigation error state model, and the aircraft navigation system measurement equation, and performing error correction on the aircraft navigation system to obtain accurate navigation positioning information. The present invention uses a laser beam to measure the distance to the sea surface and obtain horizon information, and uses the horizon information and inertial navigation information to perform combined filtering calculation to obtain a high-precision autonomous navigation result, thereby facilitating high-precision autonomous navigation of the aircraft in the absence of reference landmarks on the sea surface.

[0003] The above patent documents and prior art have the following technical problems when used:

[0004] Problem 1: In complex environments, such as those facing GPS denial, turbulence, or low visibility, traditional avionics signal processing methods face challenges with data processing accuracy. Traditional methods typically employ fixed sensor configurations and static filtering strategies, which struggle to adapt to environmental changes, leading to severe noise interference and reduced data quality. For example, the aforementioned patent document employs a fixed-weight Kalman filter for data fusion, which lacks dynamic adjustment capabilities and is prone to drift over long periods of operation, resulting in significant navigation errors and insufficient reliability and safety.

[0005] Problem two: Traditional avionics equipment has significant deficiencies in navigation data output and flight control coordination. Conventional signal processing methods use a static data output mode and cannot be dynamically adjusted according to user roles or flight scenarios, resulting in increased cognitive load on pilots and prolonged reaction time. Traditional methods lack the ability to predict control needs, and the feedback mechanism is mostly passive response. Especially in autonomous navigation of drones or low-visibility scenarios of civil aviation, poor coordination between navigation and control leads to low flight efficiency and increased safety risks, limiting the human-computer interaction experience. Summary of the Invention

[0006] Technical problems solved

[0007] In view of the shortcomings of the existing technology, the present invention provides an avionics equipment and signal processing method based on high-precision inertial navigation, which solves the following problems:

[0008] 1. Address the problem of insufficient accuracy in multi-source data processing in complex environments, resulting in large errors, insufficient reliability and security;

[0009] 2. Address the problem of poor coordination between navigation data output and flight control, which leads to low flight efficiency and poor human-computer interaction experience.

[0010] Technical Solution

[0011] To achieve the above objectives, the present invention is implemented through the following technical solutions: an avionics device based on high-precision inertial navigation and a signal processing method, the signal processing method comprising the following steps:

[0012] Sp1: Through dynamic multimodal collaborative acquisition, the sensor combination is selected according to the environmental dynamic index to collect high-precision inertial data and optical data, and data synchronization is achieved through a high-precision time protocol, providing a high-quality multi-source data foundation for subsequent signal processing;

[0013] Sp2: Based on the synchronized data collected by Sp1, through predictive acquisition triggering, analyzes historical angular velocity and acceleration data, predicts key flight events, and adjusts sampling parameters in advance to capture high-value data, thereby optimizing the pertinence and efficiency of data collection;

[0014] Sp3: Leveraging the high-value data captured by Sp2, adaptive signal cleaning is performed to dynamically adjust filtering strategies based on environmental labels, removing complex noise while retaining key signal features, providing clean input signals for feature enhancement and navigation solutions.

[0015] Sp4: Based on the signal cleaned by Sp3, multi-dimensional feature enhancement is used, and the feature reconstruction mapping method is used to extract and enhance the spatiotemporal features of inertial and optical data, generating high-dimensional feature vectors to provide rich feature support for subsequent navigation solutions;

[0016] Sp5: Combined with the high-dimensional feature vector generated by Sp4, the attitude, velocity, and position are calculated based on the quaternion method through inertial navigation solution. The direction and distance of the aircraft relative to the target are calculated using optical data through the line of sight navigation module to generate the preliminary navigation state.

[0017] Sp6: Based on the preliminary navigation status of Sp5, through dynamic multi-source fusion and environmental adaptive optimization, dynamically adjust the fusion weight of inertial and line of sight data, optimize navigation solution parameters, and apply predictive error to correct long-term drift to generate high-precision navigation results;

[0018] Sp7: Utilizes the high-precision navigation results generated by Sp6, dynamically outputs navigation data based on user roles and flight scenarios through context-adaptive visualization and predictive interactive feedback, and predicts control requirements to provide active feedback, completing the coordinated optimization of navigation and flight control.

[0019] Preferably, the dynamic multimodal collaborative acquisition in Sp1 includes the following steps:

[0020] Sp1.1: Use air pressure, temperature and humidity sensors to calculate environmental dynamic index and assess environmental complexity;

[0021] Sp1.2: Select the sensor combination based on the environmental dynamic index. LiDAR and high-frequency inertial sensors are used in high-dynamic scenarios, while spectral imaging and infrared sensors are used in low-visibility scenarios.

[0022] Sp1.3: Synchronizes inertial and optical data through a high-precision time protocol with a synchronization error of ±0.05ms;

[0023] Sp1.4: Add environmental labels to data to support subsequent processing optimization.

[0024] Preferably, the predictive acquisition triggering of Sp2 comprises the following steps:

[0025] Sp2.1: Analyze angular velocity and acceleration data over the past 5 seconds to detect high-dynamic patterns;

[0026] Sp2.2: When the angular velocity change rate exceeds the threshold, predict the probability of events within the next 0.5 seconds;

[0027] Sp2.3: If the probability is greater than 0.8, increase the inertial sampling rate to 600Hz and enable high-resolution optical mode;

[0028] Sp2.4: Verify the occurrence of events after collection and dynamically update the trigger threshold.

[0029] Preferably, the adaptive signal cleaning in Sp3 comprises the following steps:

[0030] Sp3.1: Perform spectral decomposition of inertial and optical data to identify high-frequency noise and low-frequency trends;

[0031] Sp3.1: Selects a cleaning strategy based on environmental labels, enhancing high-frequency filtering for high-dynamic scenes and low-frequency filtering for low-visibility scenes;

[0032] Sp3.1: Update filter parameters every 50ms based on the signal-to-noise ratio;

[0033] Sp3.1: Compare the integrity of signal characteristics before and after cleaning to ensure that key information is retained.

[0034] Preferably, the multi-dimensional feature enhancement of Sp4 comprises the following steps:

[0035] Sp4.1: Extract acceleration gradients and angular velocity trends from inertial data, and extract target boundaries and textures from optical data;

[0036] Sp4.2: Map low-quality features to high-dimensional space through feature reconstruction mapping to reconstruct high-resolution features;

[0037] Sp4.3: Fusion of inertial and optical features to generate a unified feature vector;

[0038] Sp4.4: Check the signal-to-noise ratio of the enhanced features and eliminate invalid enhancement results.

[0039] Preferably, the dynamic multi-source fusion of Sp6 comprises the following steps:

[0040] Sp6.1: Calculate the signal-to-noise ratio and consistency index of each data set to evaluate data quality;

[0041] Sp6.2: Dynamically assign fusion weights based on signal-to-noise ratio and consistency index, assigning higher weights to data with high signal-to-noise ratio;

[0042] Sp6.3: Fusion of the state vector of inertial and line-of-sight data by weighted averaging;

[0043] Sp6.4: Check the fusion result deviation and adjust the weight distribution threshold.

[0044] Preferably, the Sp5 environmental adaptive optimization uses the environmental dynamic index to monitor changes in flight status, adjusts the filter gain according to the environmental dynamic index, increases the gain to prioritize new data in high environmental dynamic index scenarios, and reduces the gain to rely on historical estimates in low environmental dynamic index scenarios, applies the adjusted gain to optimize attitude, speed and position solutions, records the adjustment effect, and optimizes future parameter settings.

[0045] Preferably, the predictive error correction in Sp5 records the error history sequence, calculates the trend stability factor, predicts the error at the next moment based on the trend stability factor, combines the historical trend and the current error, applies the predicted error to correct the navigation state, verifies the correction effect, and updates the prediction parameters.

[0046] Preferably, the context-adaptive visualization of the Sp6 identifies user roles, including pilots and engineers, through user login or device settings, and selects visualization modes according to the environmental dynamic index and mission status. Pilots are displayed with concise trajectory diagrams, engineers are displayed with detailed sensor status, the color, scale and refresh rate are dynamically adjusted, the system adapts to ambient light or vibration, user interaction behaviors are recorded, and the default visualization settings are optimized. The predictive interactive feedback of the Sp6 analyzes navigation status and error trends, predicts flight control adjustment requirements, generates visual, auditory or tactile feedback, allocates feedback channels according to urgency, feeds back the predicted adjustments to the flight control system, optimizes navigation and control coordination, records feedback effects, and adjusts prediction model parameters.

[0047] Preferably, the hardware components of the avionics equipment include:

[0048] A sensor array module, including high-precision MEMS gyroscopes, accelerometers, magnetometers, and a multimodal optical system that integrates an infrared camera, lidar, and spectral imager, supporting dynamic switching to adapt to different environmental conditions;

[0049] The data acquisition and pre-processing unit uses a combination of FPGA and DSP processors to support multi-channel parallel data acquisition and real-time signal cleaning. It is equipped with a high-precision time protocol module to ensure that the synchronization error between inertial and optical data is less than 0.05ms.

[0050] The navigation solution core module integrates an embedded GPU accelerator for performing quaternion solution, dynamic multi-source fusion and predictive error correction, and has dynamic power management functions to optimize long-duration missions;

[0051] Auxiliary system modules, including barometers, temperature and humidity sensors, and GPS receivers, adopt a modular design for easy upgrade and maintenance;

[0052] The output and control interface module supports ARINC429, MIL-STD-1553 and Ethernet protocols, and is equipped with a context-adaptive display unit to provide role-customized navigation data visualization.

[0053] Beneficial effects

[0054] The present invention provides an avionics device and a signal processing method based on high-precision inertial navigation, which has the following beneficial effects:

[0055] 1. The present invention uses a dynamic multi-source data processing method to achieve navigation accuracy in complex environments including GPS denial and turbulence, with a breakthrough from the traditional 5-10m to a position error of 0.5m and an attitude error of 0.2°. It drives sensor selection through environmental complexity assessment, optimizes the acquisition quality of inertial data such as angular velocity and acceleration and pixel coordinates, specific optical data in real time, reduces redundancy, uses spectral decomposition and adaptive filtering to dynamically clean signals for environmental noise in different frequency bands, retains key features, generates high-dimensional feature vectors through spatiotemporal feature reconstruction, enhances the expression ability of low-quality data, dynamically adjusts fusion weights based on signal-to-noise ratio and consistency index, combines predictive error correction to eliminate long-term drift, hardware support ensures processing delay of less than 10ms, and anti-interference shielding maintains data stability, breaking through the traditional navigation's dependence on external signals, achieving long-term high-precision navigation, and is particularly suitable for aviation covert missions and civil aviation flights in severe weather, improving mission success rate and safety.

[0056] 2. The intelligent data processing method of the present invention realizes the coordinated optimization of navigation and flight control, reducing the control error from 1° to 0.1° and shortening the user reaction time by 20%. It generates high-precision navigation status based on high-dimensional feature vectors and fusion results, providing reliable data for output, dynamically customizes visualization content through role and scene analysis, adjusts display parameters according to environmental changes, reduces cognitive load, uses error trend analysis to predict control needs, generates multimodal feedback of audio, video and touch, and optimizes feedback channels through priority allocation to urgently adjust the priority tactile sense. It also analyzes the feedback effect through real-time logs, dynamically updates the prediction model, and improves long-term accuracy. It breaks through the limitations of traditional static output and realizes real-time coordination of navigation and control. It is particularly suitable for autonomous navigation of drones and low-visibility scenarios in civil aviation, significantly improving flight efficiency and human-computer interaction experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A step diagram of the signal processing method of the present invention;

[0058] Figure 2 The hardware structure of the avionics equipment of the present invention;

[0059] Figure 3 This is a data transmission structure diagram of the present invention;

[0060] Figure 4 A graph showing changes in the environmental dynamic index over time according to the present invention;

[0061] Figure 5 This is a diagram of sensor combination selection of the present invention;

[0062] Figure 6 A line graph showing the predictive acquisition trigger detection angular velocity change rate and event triggering of the present invention;

[0063] Figure 7 This is a signal comparison diagram before and after signal cleaning during adaptive signal cleaning of the present invention;

[0064] Figure 8 This is a comparison diagram of the signal-to-noise ratio before and after feature enhancement during feature reconstruction mapping of the present invention;

[0065] Figure 9 The weight and error distribution diagram after dynamic fusion and error correction of the present invention;

[0066] Figure 10 This is a comparison diagram of the control errors of the predictive interactive feedback of the present invention. DETAILED DESCRIPTION

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention. Specific embodiment one:

[0069] like Figures 1 to 10 As shown, an avionics device and signal processing method based on high-precision inertial navigation achieves accurate estimation of position, velocity, and attitude through multi-modal sensor collaborative acquisition, dynamic signal processing, environment adaptive fusion, and interactive feedback mechanism. The signal processing method includes the following steps:

[0070] Sp1: Dynamic multimodal collaborative acquisition: responsible for efficiently collecting inertial data and optical data from multiple sensors, where inertial data includes angular velocity, acceleration, and magnetic field strength, and optical data includes target pixel coordinates and distance information, providing a high-quality multi-source data foundation for subsequent processing. It intelligently selects sensor combinations based on real-time environmental changes, and ensures data consistency through high-precision synchronization, optimizes acquisition efficiency, and adapts to complex flight scenarios. First, it uses air pressure, temperature, and humidity sensors to monitor environmental changes in real time and calculate the environmental dynamic index to evaluate environmental complexity. In turbulent or high-humidity environments, the environmental dynamic index increases, reflecting high-dynamic or low-visibility scenarios. The calculation of the environmental dynamic index is based on the rate of change of air pressure, temperature, and humidity, and quantifies the degree of environmental disturbance through a weighted sum of squares formula. Based on the environmental dynamic index, the system dynamically selects a sensor combination: for sharp turns or turbulent high-dynamic scenarios, lidar and high-frequency inertial sensors with a sampling rate of 500Hz are enabled to capture rapidly changing motion data. In low-visibility scenarios, including haze or at night, it switches to spectral Imagers and infrared cameras with a sampling rate of 60Hz are used to enhance target detection capabilities. Sensor selection is performed through a preset decision table, and the value of the environmental dynamic index is mapped to a specific combination to ensure targeted acquisition. Subsequently, the system synchronizes inertial and optical data through a high-precision time protocol, with the synchronization error controlled at ±0.05ms. The high-precision time protocol module uses hardware timestamps to ensure that multi-source data are aligned on the time axis, avoiding solution errors caused by sampling delays. Finally, environmental tags are added to each set of collected data, recording the environmental dynamic index value and sensor combination information. These tags provide contextual support for subsequent signal cleaning and feature enhancement, optimizing processing efficiency. The entire process is implemented through FPGA-driven parallel acquisition channels, and the data stream is transmitted to the pre-processing unit in real time to provide high-quality input for the next step. The environment-driven dynamic sensor scheduling and precise synchronization mechanism during dynamic multimodal collaborative acquisition surpass the traditional fixed sampling strategy. The collected data directly provides input for the predictive triggering of Sp2, and the environmental tags lay the foundation for signal cleaning of Sp3, ensuring the consistency of the process.

[0071] Sp2: Predictive acquisition trigger: Predictive acquisition trigger is based on the synchronous data collected by Sp1, further optimizing the pertinence and efficiency of data collection. By analyzing historical inertial data, key flight events, including sharp turns and turbulence, are predicted, and sampling parameters are adjusted in advance to capture high-value data, thereby reducing redundant acquisition and improving navigation accuracy. Predictive acquisition trigger uses the synchronous inertial data provided by Sp1 as a basis to analyze the angular velocity and acceleration time series of the past 5 seconds to detect high-dynamic patterns. The system calculates the rate of change of angular velocity and acceleration to identify abnormal fluctuations. When the rate of change exceeds the preset threshold, the system predicts the probability of events within the next 0.5 seconds. The prediction is based on trend analysis of historical data, and the frequency and amplitude of fluctuations are calculated through a sliding window. If the probability is greater than 0.8, it is judged to be an abnormal event. It is identified as a key event. Subsequently, the system adjusts the acquisition parameters: the inertial sensor sampling rate is increased to 600Hz to capture high-frequency motion details; the optical sensor enables high-resolution mode to enhance target detection accuracy. The adjustment process is triggered by hardware interrupts, and the response time is less than 1ms. After the acquisition is completed, the system verifies whether the event occurs and dynamically updates the trigger threshold based on the verification result. The entire process is closely connected with the data flow of Sp1. The collected high-value data is directly transmitted to the signal cleaning module of Sp3 to ensure the integrity of key information. Predictive acquisition triggering realizes active acquisition through predictive triggering, which is different from the passive response mechanism and improves data utilization. Its output data provides high-quality input for Sp3. The prediction results and environmental labels further support the filtering strategy selection of Sp3.

[0072] Sp3: Adaptive signal cleaning: Adaptive signal cleaning uses the high-value data captured by Sp2 to remove complex noise through dynamic filtering strategies, retain key signal features, and provide clean input signals for subsequent feature enhancement and navigation solutions. It adaptively adjusts the cleaning method according to the environmental tags generated by Sp1 to adapt to various flight scenarios; Adaptive signal cleaning is based on the inertial data and optical data transmitted by Sp2, combined with the environmental tags of Sp1, to perform spectral decomposition of the data, and use fast Fourier transform to separate high-frequency noise and low-frequency trends. Spectral analysis is performed every 100ms to ensure real-time performance. The cleaning strategy is selected according to the environmental tags: in high-dynamic scenes, high-frequency filtering is enhanced to retain rapidly changing motion features while removing ultra-high-frequency noise. In low-visibility scenes, Strong low-frequency filtering highlights the target outline and suppresses optical noise. The filtering parameters are updated every 50ms according to the signal-to-noise ratio, which is calculated by the ratio of signal power to noise power. If the signal-to-noise ratio is lower than the threshold, the system increases the filtering intensity to ensure signal quality. Finally, the feature integrity of the signal before and after cleaning is compared to check whether the key features are retained. If the loss rate exceeds 5%, the filtering parameters are adjusted and reprocessed. The cleaned signal is transmitted to Sp4 for feature enhancement. The entire process is implemented by a DSP processor. A single cleaning takes less than 2ms, ensuring real-time performance. Adaptive signal cleaning surpasses traditional static filtering methods through dynamic filtering driven by environmental tags. Its clean signal provides high-quality input for Sp4's feature enhancement, and the environmental tags continue to guide Sp4's enhancement strategy.

[0073] Sp4: Multi-dimensional feature enhancement: Multi-dimensional feature enhancement is based on the signal cleaned by Sp3, extracts and enhances the spatiotemporal features of inertial and optical data, generates high-dimensional feature vectors, and provides rich information support for the navigation solution of Sp5. The feature reconstruction mapping method is used to reconstruct low-quality features into high-resolution features to improve the availability of data in complex environments. Multi-dimensional feature enhancement uses the clean signal of Sp3 to extract acceleration gradients and angular velocity trends from inertial data, and extracts target boundaries and textures from optical data. The extraction process is performed separately for each set of data. The inertial data is updated every 100ms and the optical data is updated every 200ms. Subsequently, the low-quality features are mapped to high-dimensional space through feature reconstruction mapping. For the boundaries of blurred infrared images, the feature reconstruction mapping reconstructs high-resolution contours through spatiotemporal interpolation; for the low-signal images of inertial data, the feature reconstruction mapping reconstructs high-resolution contours through spatiotemporal interpolation. The acceleration of the noise ratio, the feature reconstruction mapping enhances the peak features through trend fitting. The feature reconstruction mapping is based on the pre-trained mapping table and optimizes the mapping parameters for different environmental labels. Next, the inertial and optical features are fused to generate a unified high-dimensional feature vector. The fusion process is achieved through vector splicing. The inertial features and optical features are combined according to the preset ratio to generate a vector with a dimension of 128. Finally, the signal-to-noise ratio of the enhanced features is checked. If the signal-to-noise ratio is lower than 15dB, the invalid enhancement results are eliminated and reprocessed. The enhanced feature vector is transmitted to Sp5 for navigation solution. The whole process is accelerated by GPU, and a single enhancement takes less than 3ms. Multi-dimensional feature enhancement realizes feature reconstruction through feature reconstruction mapping. Different from the traditional feature extraction method, its high-dimensional feature vector provides key input for Sp5 navigation solution, and the environmental label continues to support solution optimization.

[0074] Sp5: Inertial navigation solution and line-of-sight navigation: Sp5 combines the high-dimensional feature vector generated by Sp4, and generates preliminary navigation status including attitude, speed, and position through inertial navigation solution and line-of-sight navigation modules, providing basic data for the fusion and optimization of Sp6. It uses the quaternion method to accurately calculate the motion state of the aircraft and provides external reference through optical data to compensate for the long-term drift of inertial navigation. Sp5 is divided into two sub-modules: inertial navigation solution and line-of-sight navigation, which jointly generate navigation status;

[0075] Inertial navigation solution: Based on the high-dimensional feature vector of Sp4, including acceleration and angular velocity features, perform the following steps:

[0076] Attitude solution: Use quaternion method to update attitude, input angular velocity characteristics, and calculate quaternion update:

[0077]

[0078] in: is the quaternion at the current moment, describing the posture; is the quaternion at the previous moment; is the current angular velocity vector, extracted from the Sp4 eigenvector; is the time step, 0.01s;

[0079] Then normalize:

[0080]

[0081] The quaternion is updated every 10ms to generate roll, pitch and yaw angles;

[0082] Velocity and position calculation: Use acceleration characteristics to convert to the navigation coordinate system:

[0083]

[0084] Subtract gravity:

[0085]

[0086] Update speed:

[0087]

[0088] Update location:

[0089]

[0090] in, is the acceleration of the navigation coordinate system, is the current quaternion, describing the posture, is the acceleration of the body coordinate system, which is sent to Sp4 for feature vector extraction. is the inverse quaternion, The conjugate of divided by the modulus, is the acceleration after removing gravity, is the acceleration due to gravity, is the current speed, is the speed at the previous moment, is the time step, 0.02s, Current location, is the position at the previous moment; The current speed, velocity and position are updated every 20ms.

[0091] Line of sight navigation: Based on the optical features of Sp4, including target boundaries and textures, the following steps are performed:

[0092] Target detection: extract target feature points from high-dimensional feature vectors and locate the target center using boundary information;

[0093] Triangulation: Calculates the distance and angle of the aircraft relative to the target by combining the optical sensor focal length and the target pixel position;

[0094] Coordinate conversion: converts line-of-sight data to the navigation coordinate system, aligns it with the inertial data, and generates direction and distance estimates. Line-of-sight navigation is updated every 200ms to provide external reference data.

[0095] The preliminary navigation status, including attitude, speed, position, orientation, and distance, is integrated and transmitted to Sp6 for further fusion and optimization. The entire process is calculated collaboratively by FPGA and GPU, and a single solution takes less than 5ms. Sp5 enhances navigation accuracy through quaternion solution and line of sight navigation driven by high-dimensional features. Its preliminary status provides input for Sp6's fusion, and line of sight data lays the foundation for error correction.

[0096] Sp6: Dynamic multi-source fusion and optimization: Based on the preliminary navigation status of Sp5, Sp6 generates high-precision navigation results through dynamic multi-source fusion, environmental adaptive optimization and predictive error correction. It integrates inertial and line-of-sight data, dynamically optimizes solution parameters, corrects long-term drift, and ensures navigation accuracy and robustness.

[0097] When Sp6 performs dynamic multi-source fusion, it first calculates the signal-to-noise ratio and consistency index of each set of data. The signal-to-noise ratio is calculated by the signal power to noise power ratio, and the consistency index is based on the time series correlation of inertial and line-of-sight data. Then, the fusion weight is dynamically assigned based on the signal-to-noise ratio and consistency index. High signal-to-noise ratio data is assigned a high weight, while low consistency index data is assigned a lower weight. The fusion state vector is then weighted averaged:

[0098]

[0099] in, is the state vector of inertial or line of sight data, is the weight, For the Check the state vector of the fusion result. If the position deviation exceeds 0.1m or the attitude deviation exceeds 0.5°, adjust the weight threshold and re-fuse.

[0100] When Sp6 performs environmental adaptive optimization, it uses the environmental dynamic index of Sp1 to monitor changes in flight status. A high environmental dynamic index indicates turbulence or a highly dynamic scene. The filter gain is adjusted according to the environmental dynamic index. In high environmental dynamic index scenes, the gain is increased and new data is prioritized. In low environmental dynamic index scenes, the gain is reduced and historical estimates are relied upon. The adjusted gain is applied to optimize attitude, velocity, and position solutions, update the navigation status, record the adjustment effect, store the relationship between gain and error, and optimize future parameters.

[0101] When Sp6 performs predictive error correction, it records the error history sequence: ,in: , represents the difference between the observed value and the estimated value;

[0102] Calculate the trend stability factor:

[0103]

[0104] in: Trend stability factor,∈[0,∞), the smaller the value, the more stable the error trend; Error in time; Error in time; for The square of the moment error;

[0105] Predict the next moment error:

[0106]

[0107] in, is the predicted error at the next moment; is the current error; It is a trend instability factor. The larger the value, the greater the prediction dependence on the current change. is the current error change; is the tuning coefficient, the default value is 0.5, which controls the prediction step size;

[0108] Calibration status: ;

[0109] in, The state after correction, including posture, quality, position, is the current estimated state, is the prediction error; verify the correction effect, if the error reduction is less than expected, update value;

[0110] The fusion and optimization results are updated every 50ms to generate high-precision navigation results, ensuring a position error of less than 0.5m and an attitude error of less than 0.2°. These results are then transmitted to Sp7. The entire process is implemented using DSP and GPU in less than 4ms. MDPF, EAO, and PEC optimize navigation results through dynamic weighting, environmental adaptation, and error prediction, surpassing traditional fusion methods. Their high-precision results provide reliable data for Sp7's output and feedback.

[0111] Sp7: Context-Adaptive Visualization and Predictive Interactive Feedback: Sp7 leverages the high-precision navigation results of Sp6 to dynamically output navigation data and provide proactive feedback through context-adaptive visualization and predictive interactive feedback. This allows for coordinated optimization of navigation and flight control, tailoring output based on user roles and flight scenarios to enhance user experience while predicting control requirements and increasing flight stability.

[0112] When Sp7 performs context-adaptive visualization, it identifies the role through user login or device settings; it selects the visualization mode based on the Sp1 environmental dynamic index and mission status: pilots are shown a concise 3D trajectory diagram including position, speed, attitude, and large deviation alerts; engineers are shown detailed sensor status and diagnostic information including signal-to-noise ratio and error curves, dynamically adjust the color, scale, and refresh rate, adapt to ambient light or vibration, record user interactions, and optimize default settings through interaction logs;

[0113] When Sp7 performs predictive interactive feedback, it first analyzes Sp6's navigation status and error trends, predicts control adjustment needs, and then generates feedback: visual, auditory, or tactile. It allocates channels based on urgency, with tactile adjustments taking priority for emergency adjustments. It then feeds the predicted adjustments back to the flight control system, including pre-emptive adjustment of control surface angles and optimization of navigation and control coordination. Finally, it records the feedback effect. If the control error reduction is less than expected, it adjusts the prediction model parameters.

[0114] The output data is transmitted to the flight control system or display unit through the ARINC429 or MIL-STD-1553 interface with a refresh rate of 20Hz. The feedback process is triggered by real-time interrupts with a response time of less than 2ms. The entire process forms a closed loop, and the navigation results continuously optimize flight performance. Context-adaptive visualization and predictive interactive feedback improve user experience and control efficiency through context customization and predictive feedback, surpassing traditional static output. The feedback results can reversely influence Sp1's acquisition strategy to form a complete closed loop.

[0115] The system runs on an embedded avionics platform. The core hardware includes FPGA, DSP, GPU and display unit. The software is based on a real-time operating system. It manages data flow through multi-threading to ensure that the signal processing delay is less than 10ms. The system supports online updates and adapts to different aircraft. The system is adapted to a variety of scenarios: in GPS-denied environments, high-precision navigation is achieved by relying on predictive error correction; in fog or heavy rain, dynamic multimodal collaborative acquisition and multi-dimensional features enhance data quality to ensure safe flight; in long-duration missions, environmental adaptive optimization and context-adaptive visualization optimize energy consumption and user experience. Dynamic multimodal collaborative acquisition and predictive acquisition optimize data acquisition through environmental drive and event prediction, dynamic acquisition and prediction, and reduce Redundancy, improved efficiency, adaptive signal cleaning and multi-dimensional feature enhancement through dynamic filtering and feature reconstruction, adaptive signal processing, adaptation to complex environments, enhanced data availability, dynamic multi-source fusion, environmental adaptive optimization and predictive error correction to achieve dynamic fusion and error correction, ensuring high precision and robustness, context-adaptive visualization and predictive interactive feedback provide customized visualization and predictive feedback, improve user experience and control efficiency, this solution through dynamic multimodal collaborative acquisition, predictive triggering, adaptive signal cleaning, multi-dimensional feature enhancement, inertial and line of sight navigation solution, dynamic fusion and optimization, context visualization and predictive feedback, to construct a high-precision, robust and adaptable avionics equipment signal processing method. Specific embodiment two:

[0117] like Figures 1 to 10 As shown, based on the content in the above specific embodiments, the following contents are further disclosed:

[0118] The steps of the above signal processing method further include the following:

[0119] In Sp1, the environmental dynamic index is calculated to quantify environmental complexity and guide the selection of sensor combinations. The environmental dynamic index comprehensively evaluates the degree of environmental disturbance by monitoring the rate of change of air pressure, temperature, and humidity in real time. The calculation logic includes the following steps:

[0120] Data collection: barometer, accuracy of ±0.1hPa, temperature and humidity sensor, accuracy of ±0.5℃, ±2%RH, data collection every 100ms;

[0121] Change rate calculation: Calculate the change rate of the time series of air pressure, temperature and humidity data to reflect the dynamics of the environment;

[0122] Weighted fusion: Through the weighted square sum formula, each change rate is combined into a single index, and the weight is assigned according to the degree of influence of the environment on navigation;

[0123] Normalization: Normalize the calculation results to the interval [0, 1] to facilitate decision-making;

[0124]

[0125] in:

[0126] EDI: Environmental Dynamics Index, output value ∈ [0, 1], the higher the value, the more complex the environment;

[0127] : The weight of the air pressure change rate, the default value is 0.5, because air pressure changes have the greatest impact on navigation;

[0128] : The weight of the temperature change rate, the default value is 0.3. Temperature affects sensor performance;

[0129] : The weight of humidity change rate, the default value is 0.2. Humidity affects the quality of optical data;

[0130] : Air pressure change, that is, current air pressure With the previous moment The difference between

[0131] : Temperature change, that is ;

[0132] : Humidity change, that is ;

[0133] : Time interval, default is 0.1s;

[0134] : air pressure change rate, in hPa / s;

[0135] : Temperature change rate, unit is ℃ / s;

[0136] : Humidity change rate, unit is %RH / s;

[0137] : Normalization factor, the default value is 10, used to map the EDI value to [0, 1]. The maximum disturbance value is determined by experiment;

[0138] In practical applications, assuming that the pressure change rate is 0.5hPa / s, the temperature change rate is 0.2℃ / s, and the humidity change rate is 1%RH / s, the weights are 0.5, 0.3, and 0.2 respectively. ,but

[0139]

[0140] EDI≈0.058, indicating a low dynamic environment.

[0141] In Sp1, the system dynamically selects sensor combinations based on the environmental dynamic index to optimize the quality of inertial and optical data collection. Based on environmental complexity, it prioritizes sensors that provide high accuracy and adaptability to the current scenario, while also considering power consumption and data redundancy. The criteria and parameters for dynamically selecting sensor combinations are as follows:

[0142] The selection criteria are based on the environmental complexity EDI∈[0,1], which is divided into three intervals: low dynamic EDI<0.3, including smooth flight and low interference; medium dynamic 0.3≤EDI<0.7, including slight turbulence or mist; high dynamic includes sharp turns, turbulence, haze or night; sensor performance includes accuracy, sampling rate, and environmental adaptability; low power consumption combinations are preferred for power consumption selection, and distributed power management units are used to optimize energy consumption; high-frequency inertial sensor data is required for high-dynamic scenes, and high-resolution optical data is required for low visibility;

[0143] Sensor types: Inertial sensors and optical sensors are included. Inertial sensors include MEMS gyroscopes with an accuracy of ±0.01° / s and accelerometers with an accuracy of ±0.01g. Optical sensors include lidars with a distance accuracy of ±1cm, infrared cameras with a resolution of 1080p, and spectral imagers with a wavelength range of 400-1000nm. The sampling rate of the inertial sensors is 200-600Hz, the sampling rate of the optical sensors is 30-60Hz, the power consumption of the lidar is 10W, the power consumption of the infrared camera is 5W, and the power consumption of the spectral imager is 8W.

[0144] Selection process: The system calculates EDI every 100 ms and queries a preset decision table based on the EDI interval, as shown in Table 1 below. This is mapped to the sensor combination, activating the corresponding sensor and adjusting the sampling rate. Data is collected through the FPGA parallel channel with a PTP synchronization error of < 0.05 ms. For example, an EDI of 0.8 indicates high dynamics and turbulent scenarios. A 500 Hz lidar is selected to capture distance, and a 600 Hz high-frequency inertial sensor is selected to capture angular velocity. Power consumption is optimized to 15 W, making it suitable for sharp turns.

[0145] In Sp1, sensor selection is performed through a preset decision table. The decision table maps EDI values ​​to sensor combinations, taking into account environmental complexity, sensor performance, and power consumption requirements. The decision table is designed based on experimental data and aviation scenario requirements, covering low, medium, and high dynamic scenarios to ensure collection efficiency and data quality, as shown in Table 1 below:

[0146] EDI interval Environmental Scene Sensor combination Sampling rate Power consumption Application Scenario <0.3 Low dynamics: smooth flight MEMS gyroscope + accelerometer + infrared camera Gyroscope 200Hz, camera 30Hz 8 Normal cruising, clear weather 0.3-0.7 Medium dynamic: slight turbulence, mist MEMS gyroscope + accelerometer + spectral imager Gyroscope 400Hz, imager 40Hz 12 Slight turbulence, misty weather ≥0.7 High dynamic range: turbulence, haze, nighttime MEMS gyroscope + accelerometer + lidar + infrared camera Gyroscope 600Hz, LiDAR 500Hz, Camera 60Hz 15 Sharp turns, smog, and night navigation

[0147] Table 1

[0148] As shown in Table 1 above, the EDI range is divided into EDI values ​​of 0-1: low dynamic <0.3, medium dynamic 0.3-0.7, and high dynamic ≥0.7. The environmental scenarios correspond to typical aviation scenarios, including turbulence and haze. Low dynamic uses an infrared camera, which is low-power and suitable for target detection. Medium dynamic uses a spectral imager to enhance the outline of low-visibility targets. High dynamic uses a lidar + infrared camera for high-precision distance and target detection. In high-dynamic scenarios, the sampling rate is improved to capture rapidly changing data. The distributed power management unit is optimized to control the total power consumption to less than 15W, matching the actual aviation mission requirements.

[0149] Execution process: The system obtains the EDI value every 100ms, queries the decision table, matches the EDI interval, activates the sensor combination and sampling rate specified in the table, configures the FPGA acquisition channel, and transmits data in real time. An EDI of 0.5 indicates a medium-dynamic, misty environment. The decision table selects a 400Hz MEMS gyroscope and a 40Hz spectral imager with a power consumption of 12W, which are suitable for target detection in misty weather.

[0150] In Sp2, the system detects high dynamic modes through the rate of change of angular velocity and acceleration. When the rate of change exceeds the preset threshold, the prediction is triggered. The preset thresholds are: angular velocity change rate threshold: 0.1rad / s²; acceleration change rate threshold: 0.5m / s³;

[0151] Threshold setting method: Based on historical flight data, typical high-dynamic scenarios including sharp turns and turbulence angular velocity and acceleration change rates are analyzed to determine the statistical distribution of abnormal fluctuations. In a simulation environment, including a turbulence simulator, sensor responses are tested and trigger events are recorded, including the peak value of the rate of change of sharp turns. The 95% quantile is taken as the threshold. Sp2 verifies the prediction results. If the false alarm rate is >10%, the threshold is lowered by 5%; if the missed alarm rate is >10%, the threshold is increased by 5% to ensure prediction accuracy >80%. The threshold must be low enough to capture key events. The angular velocity change rate of 0.1rad / s² is for emergency turns, but to avoid false triggering. The threshold must filter out regular fluctuations including the angular velocity change rate of <0.05rad / s² in steady flight. According to the type of aircraft, UAVs and combat aircraft are adjusted. The threshold for combat aircraft is higher at 0.12rad / s², and lower at 0.08rad / s² for UAVs. The threshold calculation and comparison takes <0.5ms to meet the hardware interrupt response <1ms, such as the angular velocity change rate , trigger prediction, probability , increase the sampling rate to 600Hz to capture sharp turn data.

[0152] The feature reconstruction mapping method in Sp4 maps low-quality features to high-dimensional space through a pre-trained mapping table, reconstructs high-resolution features, and improves the usability of data in complex environments. The pre-trained mapping table is a set of parameterized mapping rules that stores the correspondence between environmental labels and feature reconstruction parameters, which is used to guide the spatiotemporal interpolation and trend fitting of FRM.

[0153] The mapping table is generated based on offline training, as shown in Table 2 below:

[0154] Environmental Label Feature Type Interpolation kernel bandwidth Fitting order Signal-to-noise ratio threshold Scene Weight turbulence acceleration gradient 0.5 2 15dB 0.6 turbulence Target Boundary 0.3 1 12dB 0.4 Low visibility Target Texture 0.8 3 18dB 0.7 Low visibility Angular velocity trend 0.4 2 15dB 0.3

[0155] Table 2

[0156] Mapping table generation process: Collect sensor data from various scenarios in a simulated environment, annotate environmental labels, extract low-quality features and high-resolution target features, use machine learning to optimize the interpolation kernel and fitting coefficients, generate mapping rules, verify the signal-to-noise ratio improvement of the reconstructed features, and store them in a solid-state storage unit for real-time query by Sp4;

[0157] Execution process: Sp4 receives the clean signal of Sp3 and the environmental label of Sp1, queries the mapping table, selects the corresponding reconstruction parameters, performs feature reconstruction mapping, reconstructs features through spatiotemporal interpolation and trend fitting, checks the reconstructed feature signal-to-noise ratio, and reprocesses it if it is <15dB. For example, in the low visibility scene, EDI=0.5, the target texture feature is blurred, the signal-to-noise ratio is 8dB, and the mapping table specifies an interpolation kernel bandwidth of 0.8 and a fitting order of 3. After reconstruction, the texture clarity is improved by 60%, and the signal-to-noise ratio reaches 18dB. The mapping table improves the feature reconstruction efficiency by 30%, and the target detection accuracy is increased from 70% to 95%, providing high-quality input for Sp5.

[0158] In Sp4, inertial features and optical features are fused through vector splicing to generate a unified high-dimensional feature vector, providing rich information for Sp5's navigation solution. During fusion, inertial and optical features are combined according to a preset ratio. The ratio is dynamically determined based on the environmental label and data quality to ensure the representativeness and robustness of the fused features. The preset ratio is determined by the signal-to-noise ratio of the environmental label and features from Sp1. The default ratio is as follows:

[0159] In highly dynamic scenes, including turbulence, EDI ≥ 0.7, with an inertial signature and optical signature ratio of 0.6:0.4. In turbulent environments, inertial data is more reliable, has a high signal-to-noise ratio, and prioritizes capturing motion features.

[0160] In low-visibility scenarios, including fog and haze, at night, with EDI ≥ 0.3 and high humidity, the inertial signature: optical signature ratio is 0.3:0.7. Optical data is more critical for target detection, and the signal-to-noise ratio reaches 18dB after reconstruction, prioritizing target feature enhancement.

[0161] Moderately dynamic scenes, including mild turbulence, 0.3≤EDI<0.7, inertial signature: optical signature = 0.5:0.5; inertial and optical data have similar signal-to-noise ratios, allowing for balanced fusion;

[0162] Combination parameters: Vector dimension: 64-dimensional inertial features, including 32-dimensional acceleration gradient, 32-dimensional angular velocity trend, 64-dimensional optical features, including 32-dimensional target boundary, 32-dimensional texture, 128-dimensional after fusion; Signal-to-noise ratio threshold: Inertial feature signal-to-noise ratio > 15dB, optical feature signal-to-noise ratio > 12dB, otherwise adjust the ratio;

[0163] Fusion formula:

[0164]

[0165] in, is the fusion vector, is the inertial eigenvector, is the optical eigenvector, is the inertial characteristic ratio, ranging from 0.3 to 0.6;

[0166] Combination process: Sp4 receives the clean signal of Sp3, extracts inertial and optical features, calculates the feature signal-to-noise ratio, combines the Sp1 environment label to determine the ratio, and performs weighted fusion by vector splicing: Inertial feature: 64-dimensional vector multiplied by ; Optical characteristics: 64-dimensional vector multiplied by Generate a 128-dimensional feature vector and check if the signal-to-noise ratio is >15dB. If not, adjust the ratio and re-integrate it, and transmit it to Sp5 for navigation solution. For turbulent scenes with EDI=0.8, inertial feature signal-to-noise ratio=20dB, and optical feature signal-to-noise ratio=10dB, select a ratio of 0.6:0.4. After fusion, the feature vector signal-to-noise ratio is 18dB, and the target detection accuracy is improved to 90%. Dynamic ratio fusion improves the robustness of the feature vector by 30%, and the navigation solution error is reduced from 1m to 0.5m, adapting to a variety of environments. Specific embodiment three:

[0168] like Figures 1 to 10 As shown, based on the content in the above specific embodiments, the following contents are further disclosed:

[0169] According to the signal processing method of the first embodiment, the corresponding hardware components of the avionics equipment further include the following:

[0170] Sensor array module: It is the starting point of the signal processing flow and is responsible for the data acquisition tasks of Sp1 and Sp2. It switches multimodal sensors based on the environmental dynamic index, combines high-precision MEMS gyroscopes, accelerometers, magnetometers and multimodal optical systems. The multimodal optical system includes infrared cameras, lidars, and spectral imagers to achieve accurate collection of inertial and optical data. After the module is initialized, the air pressure and temperature and humidity sensors calculate the environmental dynamic index in real time to reflect the complexity of the environment. According to the environmental dynamic index, the system selects the sensor combination through pre-programmed decision logic, enables lidars and high-frequency inertial sensors in high-dynamic scenes, and switches to spectral imagers and infrared cameras in low-visibility scenes. Data is transmitted to the data acquisition unit via an internal high-speed bus. The acquisition frequency is dynamically adjusted based on the predictive acquisition trigger of Sp2. The module is made of anti-vibration material and supports a temperature range of -40°C to +85°C, ensuring stability at high altitudes or inclement weather. The sensor array module provides high-quality multi-source data for Sp1 and Sp2 through dynamic switching and precise acquisition. The data synchronization error is less than 0.05ms, significantly improving the accuracy of subsequent signal processing. Its environmental adaptability ensures that the system can still capture reliable data in GPS-denied or low-visibility scenarios. In foggy and hazy environments, the spectral imager can clearly identify the target outline and support Sp5's line-of-sight navigation, improving the overall navigation accuracy to a position error of less than 0.5m.

[0171] Data acquisition and preprocessing unit: The data acquisition and preprocessing unit is the core execution hardware of Sp1, Sp2 and Sp3. It adopts FPGA and DSP combination processor, supports multi-channel parallel data acquisition and real-time signal cleaning, synchronizes sensor data through high-precision time protocol, and performs preliminary filtering to provide clean signals for Sp3. The data acquisition and preprocessing unit receives inertial data and optical data from the sensor array, and realizes parallel acquisition through the multi-channel ADC of FPGA. The high-precision time protocol module uses hardware timestamp to ensure that the synchronization error is less than 0.05ms, meeting the synchronization requirements of Sp1. The collected data is dynamically triggered according to the predictive acquisition instruction of Sp2. Adjustment, then, DSP performs the adaptive signal cleaning task of Sp3, identifies noise through spectral decomposition, and selects the filtering strategy according to the environmental label from Sp1, enhances high-frequency filtering in high-dynamic scenes, and updates the filtering parameters every 50ms according to the signal-to-noise ratio. The cleaned signal is stored in the buffer and transmitted to Sp4. The high-speed parallel acquisition and real-time cleaning capabilities of the unit ensure the efficient execution of Sp1 to Sp3. After signal cleaning, the signal-to-noise ratio is improved to more than 15dB, providing high-quality input for the feature enhancement of Sp4. Its low-latency processing supports real-time navigation, especially in high-dynamic scenes. The cleaned signal retains key motion features and significantly reduces the solution error of Sp5.

[0172] Navigation solution core module: It is the computing core of Sp5 and Sp6, integrating embedded GPU accelerator, performing quaternion solution, dynamic fusion and error correction, using GPU parallel computing capability to process high-dimensional feature vectors, generating high-precision navigation results, and optimizing long-duration missions through dynamic power consumption management. The navigation solution core module receives the high-dimensional feature vectors of Sp4, and the GPU executes the quaternion solution of Sp5 through the CUDA kernel, updating the attitude, speed and position. At the same time, the sight navigation submodule processes optical features and calculates the target direction and distance. In Sp6, the module performs dynamic multi-source fusion, calculates the fusion weight through the signal-to-noise ratio and consistency index, and fuses Combined inertial and line of sight data, environmental adaptive optimization adjusts the filter gain according to the environmental dynamic index, optimizes the solution parameters, predictive error correction predicts the error through the trend stability factor, and corrects long-term drift. The power management unit adjusts the GPU frequency according to the task load. For example, it reduces the power to 50% in low-dynamic scenarios to extend the flight time. The navigation solution core module uses GPU acceleration and high-precision algorithms to ensure that the navigation results of Sp5 and Sp6 achieve a position error of less than 0.5m and an attitude error of less than 0.2°. Its dynamic power consumption management reduces energy consumption by 20%, which is suitable for long-duration UAV missions. The redundant dual FPGA design ensures fault tolerance and enhances system reliability.

[0173] Auxiliary system module: supports the environmental perception functions of Sp1 and Sp6, including a barometer, temperature and humidity sensors, and a GPS receiver. It adopts a modular design for easy upgrades, provides environmental reference data for the system, and optimizes sensor selection and solution parameters. In Sp1, the barometer and temperature and humidity sensors collect data every 100ms, calculate the environmental dynamic index, and guide the selection of sensor combinations for dynamic multimodal collaborative acquisition. The GPS receiver provides initial position calibration during initialization and provides auxiliary correction data for Sp6 when GPS is available. In the environmental adaptive optimization of Sp6, the environmental dynamic index is used to adjust the filter gain. The module communicates with the main system through the I2C bus and supports hot-swappable upgrades. The data is stored in a local buffer for use by Sp3 and Sp6. The auxiliary system module improves the adaptability of Sp1 and Sp6 through precise environmental perception. For example, in turbulent environments, sensor switching driven by the environmental dynamic index reduces data redundancy by 10%, supporting rapid maintenance. GPS-assisted correction reduces the initial positioning error from 5m to 2m, enhancing the stability of the system in various scenarios.

[0174] The Output and Control Interface Module (OCI) is responsible for the Sp7's navigation data output and feedback. It supports ARINC 429, MIL-STD-1553, and Ethernet protocols and is equipped with a context-adaptive display unit. It dynamically outputs navigation results based on the user's role and flight scenario, and provides predictive feedback. The OCI receives high-precision navigation results from the Sp6 and transmits them to the flight control system via an ARINC 429 interface. In context-adaptive visualization, the display unit selects a visualization mode based on the user's role: pilots display a 3D trajectory graph, while engineers display a sensor signal-to-noise ratio curve. Display parameters are dynamically adjusted based on the environmental dynamics index. In predictive interactive feedback, the module analyzes navigation status, predicts control requirements, and generates visual, auditory, or tactile feedback. High-priority feedback is transmitted in real time via MIL-STD-1553. Feedback results are logged and used to optimize the prediction model. Customized visualization and predictive feedback enhance the Sp7's user experience and control efficiency. Multi-protocol support ensures compatibility with existing aviation systems. Real-time feedback reduces control error to as little as 0.1°, enhancing flight stability.

[0175] The anti-interference shielding module protects the sensor array and data acquisition unit from high-intensity electromagnetic interference. Utilizing electromagnetic shielding materials and a multi-layer circuit board design, it supports stable data acquisition for Sp1 through Sp3, ensuring data quality through physical shielding and signal isolation. At the hardware level, the anti-interference shielding module utilizes high-conductivity shielding material covering the sensor and acquisition circuitry, shielding against EMI from 10kHz to 1GHz. The multi-layer circuit board utilizes a ground layer to isolate high-frequency interference, and differential signal transmission is used to reduce noise coupling. In Sp1 and Sp2, the shielding module ensures the signal-to-noise ratio (SNR) of the sensor array's raw data remains above 20dB. In Sp3, isolation circuitry protects the DSP's filtering process, preventing external interference from affecting spectral decomposition. The module monitors interference intensity in real time and, if it exceeds a threshold, triggers an alarm and enhances shielding. The anti-interference shielding module significantly improves signal stability for Sp1 through Sp3. For example, in strong electromagnetic interference environments, the packet loss rate is reduced from 10% to 1%, ensuring reliable operation of the sensor array under harsh conditions and supporting the system's high reliability in combat aviation.

[0176] The Distributed Power Management Unit (DPMU) optimizes power consumption for sensors, processors, and displays through dynamic power allocation based on task priority, supporting low-energy operation across the entire signal processing pipeline. Power allocation is dynamically adjusted based on task load, extending flight time. During initialization, the DPMU loads a task priority table, prioritizing computation tasks in Sp5 and Sp6 over display tasks in Sp7. During high-load acquisition tasks in Sp1 and Sp2, 80% of power is allocated to the sensors and FPGA. During preprocessing tasks in Sp3 and Sp4, the DSP and GPU power allocation is increased to 60%. In low-dynamic scenarios, the unit reduces sensor sampling rates and GPU frequency, saving 30%. Power allocation is adjusted in real time by the PMIC chip, and power consumption data is stored in a log to optimize future allocation strategies. The entire process operates over a temperature range of -40°C to +85°C, reducing overall energy consumption by over 20%. For example, the flight time of a long-endurance drone can be extended from 8 hours to 10 hours. This dynamic allocation ensures stability during high-load missions while reducing standby power consumption, meeting the energy-saving needs of both civil and combat aviation.

[0177] Redundant backup mechanism: Dual FPGA processors are integrated into the navigation solution core module to support high-reliability operation of Sp5 and Sp6, ensuring navigation continuity when the main processor fails. The main processor status is monitored through heartbeat signals and automatically switched to the backup processor. The main FPGA performs quaternion solution of Sp5 and fusion tasks of Sp6, sending heartbeat signals to the backup FPGA every 10ms. If the main FPGA overheats or fails, the backup FPGA takes over the task within 1ms and loads the latest status data from the buffer of Sp6. During the switching process, the system maintains navigation output continuity with an error increment of less than 0.1m. The backup FPGA regularly synchronizes the calculation parameters of the main FPGA to ensure seamless switching. Fault logs are stored in the storage unit of Sp7 for subsequent analysis, reducing the system failure rate from 1% to 0.01%. The switching time is less than 1ms, ensuring navigation continuity of Sp5 and Sp6. In the GPS-denied environment of combat aviation, the system can still maintain high-precision navigation. Its high reliability supports the continuous operation of critical missions.

[0178] Environmentally adaptable hardware design: Low-temperature and vibration-resistant materials support the sensor array and data acquisition unit, adapting to the harsh environments of Sp1 to Sp3. High-toughness materials and thermal management technologies ensure hardware stability under extreme conditions. The sensor array uses ceramic-encapsulated MEMS sensors and vibration-resistant optical lenses, capable of withstanding 10g vibration and temperatures ranging from -40°C to +85°C. The data acquisition unit utilizes an aircraft-grade aluminum alloy housing with a built-in heat pipe heatsink to maintain stable operation of the FPGA and DSP at high altitudes. In Sp1, the material protects the sensor from turbulent vibration. In Sp2 and Sp3, the thermal management system ensures the accuracy of the ADC and filtering circuits, monitors temperature and vibration in real time, and triggers frequency reduction protection if thresholds are exceeded, prioritizing core data acquisition tasks and ensuring system reliability during polar flight or high-altitude missions. For example, in a high-altitude environment of -30°C, sensor data accuracy remains at ±0.01° / s, supporting stable operation of Sp1 to Sp3. Its vibration resistance reduces data packet loss by 5%, enhancing civil aviation safety in adverse weather.

[0179] High-reliability data storage unit: equipped with solid-state memory, supports backup and recovery of key data of Sp3, Sp6 and Sp7, ensures data integrity, supports system fault tolerance and analysis through real-time backup and fast recovery, and the high-reliability data storage unit receives the cleaning signal of Sp3, the error history sequence of Sp6 and the feedback log of Sp7, writes to the solid-state memory every 100ms, and uses RAID-1 mirror backup to prevent data loss. In the event of system failure or restart, the storage unit restores the latest data within 500ms and supports the PEC fast reconstruction status of Sp6. The memory protects data security through encryption chip to prevent unauthorized access. Data is compressed regularly to free up space to support long-duration missions and ensure 99.99% integrity of key data. For example, in long-duration UAV missions, 256GB capacity supports 30 days of continuous recording, and fast recovery reduces the system restart time to 1 second. Its data protection function meets the confidentiality requirements of combat aviation and supports long-term optimization analysis of Sp6 and Sp7. Specific embodiment four:

[0181] like Figures 1 to 10 As shown, based on the content in the above specific embodiments, the following contents are further disclosed:

[0182] According to the content of the above specific embodiment, further combined with three application environments, GPS denial, bad weather, and low visibility scenarios, further include the following contents:

[0183] High-precision navigation in GPS-denied environments: In GPS-denied environments for combat aviation, including electronic warfare or enemy jamming, traditional navigation systems lack external signal correction, resulting in errors rapidly accumulating to 5-10 meters, seriously impacting the execution of covert missions. This technical solution ensures high-precision navigation through dynamic multi-source data processing using signal processing methods, making it particularly suitable for covert reconnaissance missions. Specifically, it includes the following steps:

[0184] Sp1: The dynamic multimodal collaborative acquisition system uses air pressure and temperature and humidity sensors to calculate the environmental dynamic index. In high-interference environments, the environmental dynamic index is high. LiDAR and high-frequency inertial sensors with a sampling rate of 500Hz are used to collect high-precision angular velocity and distance data. The angular velocity error is ±0.01° / s, and the distance data error is ±1cm. The high-precision time protocol ensures synchronization error of less than 0.05ms, providing high-quality data for subsequent processing.

[0185] Sp2: Predictive acquisition triggers analysis of 5 seconds of historical data to predict sharp turns caused by interference with a probability greater than 0.8. This increases the inertial sampling rate to 600Hz, capturing key motion features, optimizing data targeting, and reducing redundancy by 30%.

[0186] Sp3: Adaptive signal cleaning removes high-frequency electromagnetic interference noise above 50 Hz through spectral decomposition. It uses a 10-50 Hz high-frequency filter and updates parameters every 50 ms based on the signal-to-noise ratio. After cleaning, the signal-to-noise ratio is improved to 20 dB, preserving motion characteristics.

[0187] Sp4: Multi-dimensional feature enhancement extracts acceleration gradients and target boundaries, enhances low signal-to-noise ratio optical data through feature reconstruction mapping, generates 128-dimensional feature vectors, and improves target expression capabilities;

[0188] Sp5: Inertial and line-of-sight navigation solution uses quaternion method to solve attitude and position, with attitude error less than 0.2° and position error less than 0.5m. Combined with lidar data, the target orientation is calculated to generate a preliminary navigation state;

[0189] Sp6: Dynamic multi-source fusion and optimization dynamically allocates fusion weights based on signal-to-noise ratio and consistency index. The lidar weight is 0.7. The filter gain is adjusted through environmental adaptive optimization. Predictive error correction eliminates drift and reduces the navigation error to 0.5m.

[0190] Sp7: Context-adaptive visualization and predictive interactive feedback provide pilots with a concise 3D trajectory map, predicting yaw adjustment needs, and feedback latency of less than 2ms, supporting stealthy navigation decisions.

[0191] index Traditional methods This application method Improvement Position error 5-10m 0.5m 90%-95% Attitude error 1-2° 0.2° 80%-90% Data synchronization error 0.1ms 0.05ms 50% Signal-to-noise ratio 10dB 20dB 100% Data redundancy rate 50% 20% 60%

[0192] Table 3

[0193] According to Table 3 above, through dynamic data processing from Sp1 to Sp6, the system achieves long-term high-precision navigation in a GPS-denied environment. The anti-interference shielding module ensures data stability. The feedback from Sp7 optimizes decision-making efficiency and increases the success rate of covert missions by 30%.

[0194] Ensuring safe flight in adverse weather conditions: Civil aviation faces challenges such as low visibility and turbulence in adverse weather conditions, including fog and haze, and heavy rain. Traditional navigation systems suffer from positioning errors of up to 5 meters due to noise interference and signal attenuation, increasing landing risks. This technical solution ensures navigation accuracy and flight safety through signal processing methods and environmentally adaptable hardware. The specific steps are as follows:

[0195] Sp1: Due to the high environmental dynamic index in haze environments, dynamic multimodal collaborative acquisition is used. Based on the environmental dynamic index, the spectral imager and infrared camera are switched. The spectral imager has a wavelength of 400-1000nm, and the infrared camera has a resolution of 1080p. Target contour data is collected with a high-precision time protocol synchronization error of less than 0.05ms.

[0196] Sp2: Predictive acquisition triggers predicted turbulence events, i.e., when the angular velocity change rate is greater than 0.1 rad / s², increasing the infrared camera frame rate to 60 Hz, capturing high-value target data and reducing data loss by 20%;

[0197] Sp3: Adaptive signal cleaning uses low-frequency filtering with a cutoff frequency of 5Hz to remove speckle noise. After cleaning, the clarity of target boundaries is improved by 50%, and the signal-to-noise ratio reaches 15dB.

[0198] Sp4: Multi-dimensional feature enhancement reconstructs the boundary features of blurred infrared images, generates high-dimensional feature vectors, and improves target detection accuracy to 95%;

[0199] Sp5: Inertial and line-of-sight navigation solvers calculate attitude and position with an error of less than 0.5m. Infrared data is used to calculate runway distance with an accuracy of ±0.5m, supporting precise landing.

[0200] Sp6: Dynamic multi-source fusion and optimization dynamically allocates infrared data with a high weight of 0.6, and environmental adaptive optimization increases the filter gain to 0.8 at high center-to-center ratios, corrects predictive error drift, and maintains a navigation error of 0.5m;

[0201] Sp7: Context-adaptive visualization and predictive interactive feedback provide pilots with a runway trajectory map and predictive landing adjustments. Feedback is transmitted through the tactile channel with a latency of less than 2ms, reducing cognitive load.

[0202] index Traditional methods This application method Improvement Positioning error 5m 0.5m 90% Object detection accuracy 70% 95% 36% Data loss rate 30% 10% 67% Signal-to-noise ratio 8dB 15dB 87.5% Landing collision risk 10% 3% 70%

[0203] Table 4

[0204] As shown in Table 4 above, Sp1 to Sp6 ensure navigation accuracy in severe weather conditions through dynamic processing and environmental adaptability design. The hardware includes anti-vibration materials and redundant backups to support system stability. The feedback of Sp7 improves landing safety and reduces collision risk by 70%.

[0205] Autonomous UAV navigation in low-visibility scenarios: UAVs in low-visibility scenarios, including nighttime forest patrols, require highly autonomous navigation. Traditional systems, due to static output and passive feedback, have control errors of up to 1° and response times of 2-3 seconds, limiting mission efficiency. This technical solution achieves efficient autonomous navigation through intelligent data processing. The specific steps are as follows:

[0206] Sp1: Dynamic multimodal collaborative acquisition selects an infrared camera and a low-frequency inertial sensor with a sampling rate of 200Hz to collect nighttime target data with a synchronization error of less than 0.05ms;

[0207] Sp2: Predictive acquisition triggers predicted obstacle approach events, increases the infrared frame rate to 60Hz, captures target textures, and improves data validity by 40%;

[0208] Sp3: Adaptive signal cleaning and low-frequency filtering removes nighttime speckle noise. After cleaning, the signal-to-noise ratio reaches 18dB and the target feature retention rate is 95%;

[0209] Sp4: Multi-dimensional feature enhancement reconstructs target texture features and generates high-dimensional feature vectors, achieving a target recognition accuracy of 90%;

[0210] Sp5: Inertial and line-of-sight navigation solvers calculate the attitude and position of the drone, ensuring that the attitude error is less than 0.2° and the position error is less than 0.5m. The obstacle distance is calculated with an accuracy of ±0.3m.

[0211] Sp6: Dynamic multi-source fusion and optimization dynamically assigns infrared data a weight of 0.65, environmental adaptive optimization optimizes the filter gain, predictive error correction corrects the error, and navigation accuracy is maintained at 0.5m;

[0212] Sp7: Context-adaptive visualization and predictive interactive feedback provide the operator with an obstacle avoidance trajectory map and predictive path adjustments. The tactile feedback delay is less than 2ms, reducing reaction time by 20%.

[0213] index Traditional methods This application method Improvement Control error 1° 0.1° 90% Reaction time 2-3s 1.6s 20%-47% Target recognition accuracy 60% 90% 50% Data Validity 50% 90% 80% Task completion efficiency 60% 90% 50%

[0214] Table 5

[0215] As shown in Table 5 above, the intelligent processing of Sp1 to Sp6 ensures the navigation accuracy of the drone in low-visibility scenarios. The predictive feedback of Sp7 improves obstacle avoidance efficiency. The hardware includes solid-state storage and power management, supports long-flight missions, and improves mission completion efficiency by 50%.

[0216] This technical solution solves the problems of insufficient multi-source data processing accuracy and poor coordination between navigation and control in complex environments through seven steps of signal processing methods. Specific summary data is shown in Table 6 below:

[0217] Application Scenario Core indicators Traditional methods This application method Improvement GPS denial (combat use) Position error 5-10m 0.5m 90%-95% Inclement Weather (Civil Aviation) Landing collision risk 10% 3% 70% Low visibility (drone) Task completion efficiency 60% 90% 50%

[0218] Table 6

[0219] As shown in Table 6 above, high-precision navigation ensures a position error of 0.5m and an attitude error of 0.2°. It is adaptable to GPS denial and severe weather scenarios. During control collaborative optimization, the control error is 0.1° and the reaction time is shortened by 20%. The efficiency of autonomous navigation of UAVs is improved, including dynamic multi-source data processing and intelligent data processing. Dynamic multi-source data processing includes environment-driven acquisition, adaptive cleaning, feature reconstruction, dynamic fusion, and error correction. Intelligent data processing includes customized visualization, predictive feedback, and log optimization. Hardware support ensures system reliability. This solution significantly improves safety, efficiency, and autonomy in combat aviation, civil aviation, and UAV navigation.

[0220] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further restrictions, an element defined by the statement "comprising a reference structure" does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0221] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A signal processing method for avionics equipment based on high-precision inertial navigation, characterized in that: The signal processing method comprises the following steps: Sp1: Using dynamic multimodal collaborative acquisition, the sensor combination is selected based on the environmental dynamic index to collect high-precision inertial and optical data, and the data is synchronized through a high-precision time protocol; Sp2: Based on the synchronized data collected by Sp1, it uses predictive acquisition triggering to analyze historical angular velocity and acceleration data, predict key flight events, and adjust sampling parameters in advance to capture high-value data; Sp3: Leveraging the high-value data captured by Sp2, it uses adaptive signal cleaning to dynamically adjust filtering strategies based on environmental labels, remove complex noise, retain key signal features, and provide clean input signals for feature enhancement and navigation solutions. Sp4: Based on the signal cleaned by Sp3, multi-dimensional feature enhancement is used, and the feature reconstruction mapping method is used to extract and enhance the spatiotemporal features of inertial and optical data to generate high-dimensional feature vectors; Sp5: Combined with the high-dimensional feature vector generated by Sp4, the inertial navigation solution is used to calculate the attitude, velocity and position based on the quaternion method. The line-of-sight navigation module uses optical data to calculate the direction and distance of the aircraft relative to the target to generate a preliminary navigation state. Sp6: Based on the preliminary navigation status of Sp5, it uses dynamic multi-source fusion and environmental adaptive optimization to dynamically adjust the fusion weights of inertial and line-of-sight data, optimize navigation solution parameters, and apply predictive error to correct long-term drift to generate high-precision navigation results; Sp7: Leveraging the high-precision navigation results generated by Sp6, using context-adaptive visualization and predictive interactive feedback, it dynamically outputs navigation data based on user roles and flight scenarios, and provides proactive feedback based on predicted control requirements, completing the coordinated optimization of navigation and flight control.

2. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The dynamic multimodal collaborative acquisition in Sp1 includes the following steps: Sp1.1: Use air pressure, temperature and humidity sensors to calculate environmental dynamic index and assess environmental complexity; Sp1.2: Select the sensor combination based on the environmental dynamic index. LiDAR and high-frequency inertial sensors are used in high-dynamic scenarios, while spectral imaging and infrared cameras are used in low-visibility scenarios. Sp1.3: Synchronizes inertial and optical data through a high-precision time protocol with a synchronization error of ±0.05ms; Sp1.4: Add environmental labels to data to support subsequent processing optimization.

3. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The predictive acquisition triggering of Sp2 includes the following steps: Sp2.1: Analyze angular velocity and acceleration data over the past 5 seconds to detect high-dynamic patterns; Sp2.2: When the angular velocity change rate exceeds the threshold, predict the probability of events within the next 0.5 seconds; Sp2.3: If the probability is greater than 0.8, increase the inertial sampling rate to 600Hz and enable high-resolution optical mode; Sp2.4: Verify the occurrence of events after collection and dynamically update the trigger threshold.

4. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The adaptive signal cleaning in Sp3 includes the following steps: Sp3.1: Perform spectral decomposition of inertial and optical data to identify high-frequency noise and low-frequency trends; Sp3.1: Selects a cleaning strategy based on environmental labels, enhancing high-frequency filtering for high-dynamic scenes and low-frequency filtering for low-visibility scenes; Sp3.1: Update filter parameters every 50ms based on the signal-to-noise ratio; Sp3.1: Compare the integrity of signal characteristics before and after cleaning to ensure that key information is retained.

5. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The multi-dimensional feature enhancement of Sp4 includes the following steps: Sp4.1: Extract acceleration gradients and angular velocity trends from inertial data, and extract target boundaries and textures from optical data; Sp4.2: Use feature reconstruction mapping to map low-quality features to high-dimensional space and reconstruct high-resolution features; Sp4.3: Fusion of inertial and optical features to generate a unified feature vector; Sp4.4: Check the signal-to-noise ratio of the enhanced features and eliminate invalid enhancement results.

6. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The dynamic multi-source fusion of Sp6 comprises the following steps: Sp6.1: Calculate the signal-to-noise ratio and consistency index of each data set to evaluate data quality; Sp6.2: Dynamically assign fusion weights based on signal-to-noise ratio and consistency index, assigning higher weights to data with high signal-to-noise ratio; Sp6.3: Fusion of the state vector of inertial and line-of-sight data by weighted averaging; Sp6.4: Check the fusion result deviation and adjust the weight distribution threshold.

7. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The Sp5 environmental adaptive optimization uses the environmental dynamic index to monitor changes in flight status, adjusts the filter gain according to the environmental dynamic index, increases the gain to prioritize new data in high environmental dynamic index scenarios, and reduces the gain to rely on historical estimates in low environmental dynamic index scenarios. The adjusted gain is applied to optimize attitude, speed and position solutions, records the adjustment effects, and optimizes future parameter settings.

8. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The predictive error correction in Sp5 records the error history sequence, calculates the trend stability factor, predicts the error at the next moment based on the trend stability factor, combines the historical trend and the current error, applies the predicted error to correct the navigation state, verifies the correction effect, and updates the prediction parameters.

9. The signal processing method for avionics equipment based on high-precision inertial navigation according to claim 1, characterized in that: The context-adaptive visualization of Sp6 identifies user roles, including pilots and engineers, through user login or device settings, and selects visualization mode according to the environmental dynamic index and mission status. Pilots are displayed with a concise trajectory diagram, engineers are displayed with detailed sensor status, and the color, scale and refresh rate are dynamically adjusted to adapt to ambient light or vibration, user interaction behavior is recorded, and default visualization settings are optimized. The predictive interactive feedback of Sp6 analyzes navigation status and error trends, predicts flight control adjustment needs, generates visual, auditory or tactile feedback, allocates feedback channels according to urgency, feeds back the predicted adjustments to the flight control system, optimizes navigation and control coordination, records feedback effects, and adjusts prediction model parameters.

10. A signal processing method for avionics equipment based on high-precision inertial navigation according to any one of claims 1 to 9, characterized in that: The hardware components of the avionics equipment corresponding to the signal processing method include: A sensor array module, including a high-precision MEMS gyroscope, accelerometer, magnetometer, and a multimodal optical system that integrates an infrared camera, lidar, and spectral imager; The data acquisition and pre-processing unit uses a combination of FPGA and DSP processors and is equipped with a high-precision time protocol module; Navigation solution core module, integrated with embedded GPU accelerator, for performing quaternion solution, dynamic multi-source fusion and predictive error correction; Auxiliary system modules, including barometer, temperature and humidity sensors, and GPS receiver; Output and control interface module with context-adaptive display unit.

Images