Intelligent multi-source autonomous navigation system

Abstract

The present invention relates to the field of navigation and guidance, and relates to an intelligent multi-source autonomous navigation system. The system comprises: a multi-source perception and measurement module, which integrates a plurality of navigation sensors and outputs observations for navigation positioning; and an information processing module, which receives the observations from the plurality of navigation sensors, and uses corresponding navigation solution algorithms to obtain navigation positioning information of a carrier, wherein a combination of a single navigation sensor and a navigation solution algorithm corresponding thereto is denoted as a single navigation information source; and according to navigation task requirements, the information processing module evaluates the navigation positioning information output by each single navigation information source to obtain an optimal navigation information source configuration scheme under a current task and a current scenario, screens out an optimal information fusion algorithm applicable to the optimal navigation information source configuration from a stored navigation algorithm library, and uses the optimal information fusion algorithm to perform multi-source fusion on outputs of the optimal navigation information source, so as to obtain an optimized navigation positioning result.

Description

A smart multi-source autonomous navigation system

[0001] This application claims priority to Chinese Patent Application No. 2024119282747, filed on December 25, 2024, entitled "An Intelligent Multi-Source Autonomous Navigation System", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This invention relates to an intelligent multi-source autonomous navigation system, belonging to the field of navigation and guidance. Background Technology

[0003] With continuous breakthroughs in key technologies such as artificial intelligence, sensor technology, communication technology, and big data processing technology, unmanned systems technology has developed rapidly. In scenarios requiring long-term, high-intensity, or high-risk operations, the application of unmanned systems can significantly reduce reliance on human labor.

[0004] Autonomy is a key characteristic of unmanned systems. To achieve autonomous perception, decision-making, and execution capabilities, high-precision positioning and navigation are essential. Currently, the navigation systems and missions of various unmanned systems are becoming increasingly complex, leading to rising demands for autonomy, safety, and reliability. Single-satellite navigation is not only susceptible to electromagnetic interference but also suffers from limitations such as obstruction, shielding, and limited transceiver equipment. Intelligent multi-source autonomous navigation systems can comprehensively utilize multiple information sources, including BeiDou, inertial, atmospheric, geomagnetic, optical, radio signals, and imagery, to determine navigation and positioning information through perception, fusion, decision-making, and evaluation. This significantly improves navigation accuracy and reliability while ensuring continuity and stability, reducing errors and uncertainties introduced by single information sources.

[0005] In existing autonomous navigation module technologies for unmanned systems, lidar is often used as the primary information source for precise autonomous navigation tasks. A robot autonomous navigation method, device, equipment, and storage medium (patent number: CN118483996A) uses lidar to acquire simultaneous localization and a cost map. This cost map is then combined with a spoken scenario to construct a multimodal semantic map, which is then integrated with the robot's multi-faceted cognitive database for autonomous navigation. While this method can effectively achieve navigation for unmanned systems, it suffers from limited information sources, poor system flexibility, and a limited range of applications.

[0006] With the increasing application scenarios and increasingly complex tasks of unmanned systems, these systems are constantly developing towards lower cost, smaller size, clustering, and larger scale. This requires their navigation systems to adopt high-density integration technology to integrate intelligent multi-source autonomous navigation systems while ensuring navigation accuracy, so that the navigation systems also need to have the characteristics of low cost and small size.

[0007] Currently, my country's unmanned systems, such as drones, unmanned vehicles, guided munitions, and loitering munitions, urgently need a low-cost, multi-functional, and cost-effective intelligent multi-source autonomous navigation system. Summary of the Invention

[0008] The technical problem solved by this invention is to provide an intelligent multi-source autonomous navigation system for unmanned systems under GNSS (Global Navigation Satellite System) denial conditions. This system achieves hardware integration of multiple information source sensors, multi-source information fusion, and autonomous navigation decision-making, thereby enhancing the system's autonomous navigation capability.

[0009] The technical solution of the present invention is: an intelligent multi-source autonomous navigation system, comprising a multi-source sensing and measurement module and an information processing module;

[0010] The multi-source sensing and measurement module integrates multiple navigation sensors and outputs observations for navigation and positioning.

[0011] The information processing module receives observations from various navigation sensors, employs corresponding navigation algorithms to obtain the carrier's navigation and positioning information, and denotes the combination of a single navigation sensor and its corresponding navigation algorithm as a single navigation information source. Based on the navigation task requirements, the module evaluates the navigation and positioning information output by each single navigation information source to obtain the optimal navigation information source configuration scheme for the current task and scenario. It then selects the optimal information fusion algorithm suitable for the optimal navigation information source configuration from the stored navigation algorithm library and uses the optimal information fusion algorithm to perform multi-source fusion on the output of the optimal navigation information source to obtain optimized navigation and positioning results.

[0012] The navigation sensors include: an IMU unit, a GNSS receiver, a visual camera, and a magnetometer;

[0013] The observations of the IMU unit include the angular velocity and acceleration information of the carrier;

[0014] The observations of the GNSS receiver include pseudorange, carrier phase and differential correction values ​​between the carrier and the navigation satellite. The differential correction values ​​include satellite clock error, receiver clock error, ionospheric delay and tropospheric delay.

[0015] The observations of a visual camera include environmental image information;

[0016] The magnetometer's observations include information on the strength and direction of the magnetic field.

[0017] The information processing module includes a navigation calculation module, an information source configuration optimization module, an availability filtering module, an optimality filtering module, an information fusion module, and a gating switch;

[0018] The navigation calculation module uses a mechanical arrangement algorithm to obtain the vehicle's velocity, position, and attitude information from the angular velocity and acceleration information output by the IMU unit; it uses a differential positioning algorithm to calculate the vehicle's position and velocity information based on the pseudorange, carrier phase, and differential correction amount output by the GNSS receiver; it uses a SLAM algorithm to calculate the vehicle's position and attitude information based on environmental image information; and it uses a MAGCOM algorithm to calculate the vehicle's position information based on the magnetic field strength and direction information.

[0019] The information source configuration optimization module acquires navigation and positioning information output from each navigation information source, calculates the quality coefficient of each navigation information source and performs normalization processing. Based on the normalized quality coefficient, which includes accuracy coefficient, real-time coefficient, anti-interference coefficient and fault tolerance coefficient, and combined with the externally input navigation source performance index weights, the optimal navigation information source configuration scheme under the current task scenario is solved through a multi-objective optimization algorithm.

[0020] The availability filtering module filters out available information fusion algorithms under the optimal navigation information source configuration;

[0021] The optimality filtering module selects the optimal information fusion algorithm from the available information fusion algorithms;

[0022] The gating switch sends the output of the configured navigation information source to the information fusion module;

[0023] The information fusion module, based on the configured navigation information source output, uses the optimal information fusion algorithm to perform multi-source fusion to obtain the fused navigation and positioning results of the carrier.

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] (1) This invention provides a novel architecture for an intelligent multi-source autonomous navigation system for unmanned systems. It encapsulates the information processing module, power management module, multi-source sensing and measurement module, safety monitoring module, and online storage module in a metal shell. Through a "plug-and-play" open system architecture, it realizes the integrated architecture design of intelligent multi-source autonomous navigation.

[0026] (2) This invention collects multi-source sensing sensor data such as angular rate signal, acceleration signal, barometric altitude signal, magnetic intensity signal, and positioning and timing signal. Based on observability theory and mathematical statistics theory, it constructs criteria and executes multi-objective optimization algorithms in combination with task requirements to improve the reliability of intelligent multi-source autonomous navigation sensing signals in complex environments and ensure the fault tolerance capability of the sensing system in the event of sensor failure. Through multi-source fusion of the solution results of heterogeneous navigation information sources, it improves the robustness and adaptability of the navigation and positioning system.

[0027] (3) This invention provides a high-density integrated intelligent multi-source autonomous navigation system. It adopts advanced packaging technologies such as TSV, RDL, IPD, and SiP to achieve high-density integration of system modules and three-dimensional integration of the system, thereby reducing the volume, weight and power consumption of the intelligent multi-source autonomous navigation system and enabling the unmanned system to have more effective space and payload.

[0028] (4) Based on the design concept of “software-defined navigation”, this invention fully considers the universality and scalability of the hardware layer, as well as the modularity and integration of the software layer. The overall software system is divided into “navigation information calculation module”, “sensor configuration optimization module”, “availability filtering module”, “optimality filtering module” and “information fusion module” according to function, so that the overall software architecture has high readability, is easy to maintain and upgrade, and meets the needs of future military-civilian integration unmanned system autonomous and controllable technology.

[0029] (5) This invention realizes the hardware integration of multiple information source sensors such as Beidou, inertial, atmospheric, geomagnetic, optical, radio signals, and images, the fusion of multiple information sources, and autonomous guidance decision-making. It meets the functions of sensor data reception and processing, navigation calculation, multiple interface communication and AD (Analog Digital) acquisition, and solves the urgent need of various unmanned systems for "software-defined navigation" of navigation, guidance and control systems. Attached Figure Description

[0030] Figure 1 is a schematic diagram of the composition of an intelligent multi-source autonomous navigation system for unmanned systems according to a preferred embodiment of the present invention;

[0031] Figure 2 is a block diagram of a preferred embodiment of an intelligent multi-source autonomous navigation system for unmanned systems according to the present invention.

[0032] Figure 3 is a block diagram of the information processing module according to a preferred embodiment of the present invention;

[0033] Figure 4 is a flowchart of an intelligent multi-source autonomous navigation system for unmanned systems according to a preferred embodiment of the present invention.

[0034] Figure 5 is a block diagram of the principle of a software-defined navigation system for an intelligent multi-source autonomous navigation system according to a preferred embodiment of the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0036] Figure 1 is a schematic diagram of the composition of an intelligent multi-source autonomous navigation system according to a preferred embodiment of the present invention, and Figure 2 is a block diagram of the principle of an intelligent multi-source autonomous navigation system for unmanned systems according to a preferred embodiment of the present invention, including an information processing module M1, a power management module M2, a multi-source sensing and measurement module M3, a safety monitoring module M4, and an information storage module M5.

[0037] In a preferred embodiment of the present invention, the information processing module M1 consists of a main processor and a crystal oscillator. The power management module M2 receives an external input power supply voltage of 6V to 36V and converts it into a rated voltage via a DC-DC voltage converter to power the various chips inside the intelligent multi-source autonomous navigation system. Furthermore, the power management module M2 can also receive an external input power supply voltage of 5V to 36V. When the external input voltage is lower than 5V, a charge pump is required for voltage boosting and regulation.

[0038] The multi-source sensing and measurement module M3 integrates various navigation sensors and outputs observations for navigation and positioning. It consists of an IMU unit (including a micro inertial gyroscope and a micro inertial accelerometer), a GNSS receiver, a barometer, a magnetometer, a polarization sensor, a data link rangefinder, a mono / binocular vision camera, and an expansion module, among other multi-source sensing sensors.

[0039] In a preferred embodiment of the present invention, the three-axis micro gyroscope is a MEMS gyroscope, the three-axis micro accelerometer is a MEMS accelerometer, and the three-axis magnetometer is a TMR magnetometer.

[0040] The security monitoring module M4 consists of a watchdog monitoring chip. The information storage module M5 consists of FLASH memory and RAM memory.

[0041] As shown in Figure 3, the information processing module M1 communicates and transmits data with the power management module M2, the multi-source sensing and measurement module M3, the safety monitoring module M4, and the information storage module M5 via electrical connection.

[0042] The power management module M2 provides instrument power signals to the information processing module M1, the multi-source sensing and measurement module M3, the safety monitoring module M4, and the information storage module M5 via electrical connection.

[0043] Each component is encapsulated in a metal housing using a high-density integration method. The metal housing is then mounted and fixed to the unmanned system via mounting holes. This high-density integration is achieved through high-density module integration and high-density system integration. High-density module integration refers to the use of TSV adapter boards for electrical signal interconnection between chips in the information processing module M1, power management module M2, and information storage module M5; horizontal electrical signal interconnection between chips in the module using multi-layer RDL; passive device integration using IPD; electrical signal interconnection from chips in the module to the TSV adapter board using flip-chip solder balls; and the use of SiP packaging technology to achieve complete module functionality through interconnection between functional layers. High-density system integration refers to the use of rigid-flex boards for electrical signal interconnection between chips and modules; structural interconnection between the rigid-flex boards and the metal housing using thermally conductive adhesive; chip heat dissipation using thermally conductive silicone grease; and the achievement of complete intelligent multi-source autonomous navigation system functionality for unmanned systems through interconnection between functional layers.

[0044] In a preferred embodiment of the present invention, the processor in the information processing module M1 receives various sensing and feedback information input from the multi-source sensing and measurement module M3, the safety monitoring module M4, and the information storage module M5, and performs complex navigation attitude calculations. The processor in the information processing module M1 has a main frequency better than 250MHz, a computing power better than 300DMIPS, supports floating-point operations and multi-level interrupt management, and supports complex navigation attitude calculations. The crystal oscillator frequency in the information processing module M1 is better than 24MHz. The information processing module M1 is configured with a navigation interface. The navigation interface is configured with a 6V~36V power supply input, a 3.3V reference output, a power ground, 3 UART serial communication ports, 2 CAN bus ports, 1 RS232 serial communication port, 1 RS422 serial communication port, 1 SPI serial communication port, 1 I2C communication port, 1 USB communication port, 3 GPIO general-purpose interfaces, and 8 ADCs. The multi-source sensing and measurement module M3 includes a micro inertial gyroscope, a micro inertial accelerometer, a GNSS receiver, a barometer, a magnetometer, a polarization sensor, a data link rangefinder, and a mono / binocular vision camera and expansion module. It is electrically connected to the information processing module M1. Simultaneously, the power management module M2 provides power to the multi-source sensing and measurement module M3 and acts as a filter circuit to reduce ripple and surge interference. The micro inertial gyroscope is a three-axis (X / Y / Z) MEMS gyroscope, the micro inertial accelerometer is a three-axis (X / Y / Z) MEMS accelerometer, and the magnetometer is a three-axis (X / Y / Z) TMR magnetometer. The X / Y / Z axes follow the right-hand rule and are mutually orthogonal. The three-axis MEMS gyroscope outputs angular rate signals; the three-axis micro accelerometer outputs acceleration signals; the GNSS receiver outputs navigation, positioning, and timing signals; the barometer outputs barometric altitude signals; the three-axis magnetometer outputs magnetic induction intensity signals; the polarization sensor outputs navigation and positioning signals; the data link rangefinder outputs relative distance signals; and the monocular / binocular vision camera outputs image signals. The expansion module can expand to include other sensors via a plug-and-play interface protocol. The aforementioned multi-source sensing signals are transmitted to the processor of the information processing module M1 via an SPI interface, I2C interface, CAN bus, or serial port.

[0045] In a preferred embodiment of the present invention, the GNSS module can use one or any combination of BDS, GPS, Galileo, and GLONASS for positioning and timing calculation. The security monitoring module M4 monitors the status information and alarm information of each module in real time through a watchdog chip and outputs the information to the processor of the information processing module M1 via a serial port.

[0046] The information processing module receives observations from multiple navigation sensors, employs corresponding navigation calculation algorithms to obtain the carrier's navigation and positioning information, and the combination of a single navigation sensor and its corresponding navigation calculation algorithm is denoted as a single navigation information source. Based on the navigation task requirements, the module evaluates the navigation and positioning information output by each single navigation information source to obtain the optimal navigation information source configuration scheme for the current task and current scenario. It then selects the optimal information fusion algorithm suitable for the optimal navigation information source configuration from the stored navigation algorithm library and uses the optimal information fusion algorithm to perform multi-source fusion on the output of the optimal navigation information source to obtain an optimized navigation and positioning result.

[0047] The information processing module includes a navigation calculation module, an information source configuration optimization module, an availability filtering module, an optimality filtering module, an information fusion module, and a gating switch;

[0048] The navigation calculation module uses a mechanical arrangement algorithm to obtain the vehicle's velocity, position, and attitude information from the angular velocity and acceleration information output by the IMU unit; it uses a differential positioning algorithm to calculate the vehicle's position and velocity information based on the pseudorange, carrier phase, and differential correction amount output by the GNSS receiver; it uses a SLAM algorithm to calculate the vehicle's position and attitude information based on environmental image information; and it uses a MAGCOM algorithm to calculate the vehicle's position information based on the magnetic field strength and direction information.

[0049] The information source configuration optimization module acquires navigation and positioning information output from each navigation information source, calculates the quality coefficient of each navigation information source and performs normalization processing. Based on the normalized quality coefficient, which includes accuracy coefficient, real-time coefficient, anti-interference coefficient and fault tolerance coefficient, and combined with the externally input navigation source performance index weights, the optimal navigation information source configuration scheme under the current task scenario is solved through a multi-objective optimization algorithm.

[0050] The availability filtering module filters out available information fusion algorithms under the optimal navigation information source configuration;

[0051] The optimality filtering module selects the optimal information fusion algorithm from the available information fusion algorithms;

[0052] The gating switch sends the output of the configured navigation information source to the information fusion module;

[0053] The information fusion module, based on the configured navigation information source output, uses the optimal information fusion algorithm to perform multi-source fusion to obtain the fused navigation and positioning results of the carrier.

[0054] The normalization of the quality coefficients was calculated using the min-max method.

[0055] in, C is the k-th quality coefficient of the j-th navigation source; kmin C is the minimum value among all navigation sources of the k-th quality coefficient; kmax It is the maximum value among the k-th quality coefficients of all navigation sources.

[0056] The performance indicators of the navigation source include accuracy, anti-interference ability, real-time performance, and fault tolerance.

[0057] The anti-interference coefficient is calculated using the following formula: D j =a1η j +a2ε j ,

[0058] In the formula, D j η is the anti-interference coefficient of the j-th navigation information source; a1 and a2 are weighting parameters that can be adjusted based on engineering experience; j ,ε j Let be the system observability and system disturbance degree of the j-th navigation information source, respectively, calculated by the following formula:

[0059] In the formula, The state observability of the i-th output of the j-th navigation information source; Let r be the state perturbation degree of the i-th output of the j-th navigation information source; r is the number of outputs of the j-th navigation information source, i = 1 to r.

[0060] State observability It is obtained through the following method:

[0061] For a single source of navigation information, construct a linear time-invariant system: X k =FX k-1 +w k-1 Z k =HX k +υ k ,

[0062] In the formula, X k Z represents the state variables of the navigation information source, including the motion state estimation error of the vehicle and navigation information source error-related variables (such as the random walk and zero bias of the inertial navigation system's gyroscope); F is the state transition matrix; Z represents the state transition matrix. k H is the output of the navigation information source, representing the observed variable; H is the measurement matrix; w k-1 ,υ k These are process noise and measurement noise, respectively.

[0063] Construct the observability matrix Q and perform SVD singular value decomposition on it:

[0064] In the formula, Its diagonal elements are the singular values ​​of Q; U is an orthogonal matrix with column vectors QQ. T The eigenvectors of V; V is an orthogonal matrix whose column vectors are Q. T The eigenvectors of Q;

[0065] Calculate the state observability of the i-th output of the j-th navigation information source:

[0066] In the formula, σ i Let be the i-th singular value of the observability matrix of the system corresponding to the j-th navigation information source; max(σ i The value of represents the maximum value among all singular values. This index mainly reflects the characteristics of the navigation information source itself, and is determined by its measurement principle (state transition matrix F) and output (measurement matrix H).

[0067] The state perturbation It is obtained through the following method:

[0068] For the i-th output of navigation information source j Construct an extended state observer (ESO) and use its output As input to the observer, the ESO extended state observation signal of the i-th output is obtained, the variance of the ESO extended state observation signal of the i-th output is calculated, and the state perturbation degree is calculated:

[0069] In the formula, Let be the initial observation noise variance of the i-th output of the j-th navigation source. Let be the variance of the ESO extended state observation signal of the i-th output.

[0070] For the i-th output of navigation information source j Can make Build the following system:

[0071] In the formula, f(x1,x2,t) is usually an unknown term, which includes internal disturbances caused by modeling inaccuracies and external disturbances to the system; y is the system output; u is the system control input; b is the system control coefficient;

[0072] Add an extended state x3 = f(x1,x2,t) to the system. Construct an extended state observer (ESO) to observe the system states x1, x2, x3; the ESO observer takes the system control input u and the system output y as inputs and outputs... Three state observations; let Then output for navigation information source ESO observers can be constructed as follows:

[0073] In the formula, For observer output; β 01 ,β 02 ,β 03 These are the parameters of the ESO observer, and their tuning depends on engineering experience. For nonlinear feedback functions, α1 = 0.5 and α2 = 0.25 can be taken (general empirical values). δ is a constant that affects the filtering effect. The values ​​of α and δ need to be considered in conjunction with engineering experience to balance the tracking speed of the observer and the filtering effect. u is the system control input; b is the system control coefficient.

[0074] In summary, the state perturbation degree is defined as follows:

[0075] In the formula, For the i-th output of the j-th navigation information source The initial observation noise variance, Output for navigation information source The ESO extended state observation signal, i.e., the observer output. The variance.

[0076] The fault tolerance coefficient of the navigation information source is calculated using the following method:

[0077] To characterize the probability of sensor failure, a statistic can be constructed based on the redundant information of the shared output of multiple navigation sources. For the j-th navigation source, the following fault tolerance coefficient can be constructed:

[0078] In the formula, r′ represents the total number of common outputs (such as carrier position) from all navigation information sources; μ is the i-th common output of the j-th navigation information source; i The common output state z for all navigation sources i The average value.

[0079] The optimization problem for finding the optimal navigation sensor configuration is formulated as follows: f1(V)=V·[e 1 ,e 2 …e M ] T f2(V)=V·[T 1 ,T 2 …T M ] T f3(V)=V·[D 1 D 2 …D M ] T f4(V)=V·[Θ1 ,Θ 2 …Θ M ] T ,

[0080] In the formula, J is the objective function to be optimized; V is a 1×M navigation source configuration vector, M is the number of navigation sources, and the i-th column element of V is 1 to indicate that the i-th navigation source is used, and 0 to indicate that the navigation source is not used; Ω is the optimization decision space, that is, all possible navigation source configuration vectors; ω1, ω2, ω3, ω4 are the weights of the navigation source performance indicators, representing the priority of navigation source accuracy, real-time performance, anti-interference and fault tolerance in the task requirements, respectively; Here, e represents the normalized navigation source performance function using the min-max method, corresponding to the accuracy, real-time performance, anti-interference, and fault tolerance coefficients, respectively; j T represents the accuracy coefficient of the j-th navigation source, i.e., the average deviation of its common output relative to the mean of the common output of all navigation sources; j Θ is the real-time coefficient of the j-th navigation source, i.e., the time consumed in its information processing; j Let be the fault tolerance coefficient of the j-th navigation source.

[0081] Optimizing the navigation source configuration is a multi-objective optimization problem, which can be described as: min F(V)=[f1(V),f2(V)…f n (V)] T stV∈Ω Ω={V=[v1,v2…v M ]∈R M |v1,v2…v M ∈{0,1},

[0082] In the formula, F(V) is the objective function to be optimized; f1(V), f2(V)...f n (V) represents the performance index of the navigation source, and n represents the number of performance indices; Ω represents the optimization decision space, which is all possible combinations of navigation source configurations; V is a 1×M navigation source configuration vector, where M is the number of navigation sources, and the i-th column element of V is 1 to indicate that the i-th navigation source is used, and 0 to indicate that the navigation source is not used.

[0083] When solving for the optimal navigation source configuration, a weighted method can be used to transform the multi-objective optimization problem into a single-objective optimization problem, the specific form of which is as follows: f1(V)=V·[e 1 ,e 2 …e M ] T f2(V)=V·[T 1 ,T 2 …T M ] T f3(V)=V·[D1 D 2 …D M ] T f4(V)=V·[Θ 1 ,Θ 2 …Θ M ] T ,

[0084] In the formula, J is the single objective function to be optimized; ω1, ω2, ω3, and ω4 are the weights of the navigation source performance indicators, representing the priority of navigation source accuracy, real-time performance, anti-interference and fault tolerance in the task requirements, respectively. These are the navigation source performance functions normalized by the min-max method, corresponding to accuracy, real-time performance, anti-interference, and fault tolerance coefficients, respectively; e j T represents the accuracy coefficient of the j-th navigation source, i.e., the average deviation of its common output relative to the mean of the common output of all navigation sources; j Θ is the real-time coefficient of the j-th navigation source, i.e., the time consumed in its information processing; j Let be the fault tolerance coefficient of the j-th navigation source.

[0085] The obtained optimal navigation source configuration scheme V * Let M be a 1×M vector, where M is the number of navigation sources.

[0086] The specific steps of the availability filtering module are as follows:

[0087] Construct an availability matching matrix A between navigation solution algorithms and navigation sources. This availability matching matrix is ​​an M×N matrix, where M is the number of navigation sources and N is the number of navigation solution algorithms. The rows of the availability matching matrix represent navigation sources, and the columns represent navigation solution algorithms. An element of 1 in the j-th row and i'-th column indicates that the i'-th navigation solution algorithm is applicable to the j-th navigation source, and an element of 0 in the j-th row and j-th column indicates that the i'-th navigation solution algorithm is not applicable to the j-th navigation source. j = 1 to M ， i' = 1 to N;

[0088] Receive the optimal navigation source configuration vector V * The input is a 1×M vector, where the j-th element is 1 to indicate that the j-th navigation source is used, and 0 to indicate that the i'-th navigation source is discarded.

[0089] Construct a vector of available algorithms B, which is a 1×N vector. The i'th element is 1 to indicate that the i'th information fusion algorithm is available under the current navigation source configuration, and 0 to indicate that it is not available.

[0090] The navigation source configuration vector and the availability matching matrix are subjected to element-wise AND and XOR operations to obtain the available algorithm vector. The specific calculation method is: B = XNOR(V* A&V * ),

[0091] In the formula, XNOR is the element-wise XOR operation; & is the element-wise AND operation.

[0092] The output of the availability filtering module is a series of information fusion algorithm vectors decomposed from the availability algorithm vector B, and their relationship conforms to the following formula: B = P1 + P2 + ... P i ′…+P q ,

[0093] In the formula, P i′ Let P be a 1×N vector with only one element being 1 and the rest being 0. i′ A value of 1 indicates that it represents the i'th information fusion algorithm; q represents the total number of available information fusion algorithms.

[0094] The optimality screening module uses a multi-objective optimization algorithm to obtain the optimal information fusion algorithm:

[0095] The problem of finding the optimal information fusion algorithm is transformed into a single-objective optimization problem using a weighted method. The specific form of this optimization problem is as follows: f3(P)=P·[ζ 1 ,ζ 2 …ζ N ] T ,

[0096] In the formula, J′ is the objective function to be optimized; N is the number of available information fusion algorithms; P is a 1×N vector representing a specific information fusion algorithm, with only one element being 1 and the rest being 0; p i′ Ω′ represents the available information fusion algorithm; Ω′ represents the decision space, i.e., all available information fusion algorithms; ω′1, ω′2, ω′3 represent the weights of the performance indicators of the information fusion algorithms, respectively representing the priority of the information fusion algorithm in terms of accuracy, real-time performance, and anti-interference capability in the task requirements. These are the quality coefficients of the information fusion algorithm after normalization using the min-max method; is the accuracy coefficient of the j-th algorithm, which is the average offset of its fusion result relative to the fusion results of all algorithms; ζ represents the real-time performance coefficient of the j-th information fusion algorithm, i.e., the time consumed by its computation; j Let be the filtering observability of the j-th information fusion algorithm, i.e., the anti-interference coefficient;

[0097] The covariance matrix Y maintained by the information fusion algorithm can be used as a basis. k To define the filter observability:

[0098] In the formula Y 0(j′j′)Y represents the element in the j'-th row and j'-th column of the given initial covariance matrix, which is also the standard deviation of the error of the j'-th filter state. k(jj) The standard deviation of the filtered state error at time k represents the state at time k.

[0099] The optimal solution P obtained * The column number i' of the unique non-zero element indicates that the optimal information fusion algorithm is the i'th available information fusion algorithm.

[0100] The navigation solution algorithm includes extended Kalman filtering, unscented Kalman filtering, volumetric Kalman filtering, federated Kalman filtering, particle filtering, and factor graph optimization.

[0101] Extended Kalman Filter (EPF) uses a first-order Taylor expansion to approximate nonlinear systems. Particle Filter (PFF), based on Bayesian filtering, uses Monte Carlo simulations to obtain Bayesian numerical solutions. It uses a set of weighted samples to approximate the statistical distribution of the nonlinear system's state and, compared to EPF, is applicable to non-Gaussian systems. Factor graph optimization (FGM) uses a Bayesian network generative model to describe the state estimation problem, decomposing the conditional probability product of observations and state variables into factors, typically exponential functions, thus transforming the probabilistic FGM problem into a least-squares problem. Various navigation algorithms possess different computational efficiencies and accuracies, making them suitable for different task scenarios. This invention uses a multi-objective optimization algorithm to select the optimal navigation algorithm for the current task scenario from an algorithm library.

[0102] Figure 4 is a flowchart of an intelligent multi-source autonomous navigation system for unmanned systems according to a preferred embodiment of the present invention. The method includes the following steps:

[0103] Step S1: Internal self-test and initialization of the intelligent multi-source autonomous navigation system;

[0104] Step S2: Real-time monitoring of intelligent multi-source autonomous navigation status parameters and local data storage;

[0105] Step S3: Online scene switching and autonomous reconstruction of the intelligent multi-source autonomous navigation system;

[0106] Step S4: Intelligent multi-source autonomous navigation algorithm library and intelligent pattern recognition;

[0107] Step S5: Intelligent multi-source autonomous navigation and guidance task processing.

[0108] In a preferred embodiment of the present invention, step S1 includes the internal self-test of the intelligent multi-source autonomous navigation system, which includes the self-test of the underlying driver of the information processing module M1, the power management module M2, the multi-source sensing and measurement module M3, the safety monitoring module M4, and the information storage module M5, the power-on timing logic self-test, and the initialization of interrupts, clock, GPIO, I2C, ADC, SPI, UART, serial port, etc.

[0109] In a preferred embodiment of the present invention, in step S2, the real-time monitoring of the intelligent multi-source autonomous navigation status parameters is performed by the safety monitoring module M4; and the data is stored locally by the information storage module M5 mounted on the UAV.

[0110] In a preferred embodiment of the present invention, in step S3, the intelligent multi-source autonomous navigation system scenarios include reconnaissance and detection scenarios, disaster relief and firefighting scenarios, urban building scenarios, wilderness jungle scenarios, narrow street fighting scenarios, dense artillery fire scenarios, and typical extended scenarios; the intelligent multi-source autonomous navigation system for unmanned systems performs online switching of scenarios based on the information sources collected by the multi-source perception and measurement module M3 sensors, and autonomously reconstructs the working scenario based on the increase or decrease of sensor information sources.

[0111] In a preferred embodiment of the present invention, the intelligent pattern recognition function in step S4 is to intelligently identify the complexity of the navigation task based on the current scene information, thereby adaptively selecting the optimal navigation mode and realizing the upgrading and downgrading of the navigation strategy based on the task scenario. For example, under normal road conditions, the zero-speed correction algorithm can be used to improve the navigation accuracy of land vehicles, while when the vehicle is moving at high speed, the vehicle dynamics-assisted navigation method is more suitable. When the complexity of the motion environment increases further, it is necessary to use physical information neural network-assisted navigation. These three algorithms constitute a set of upgradable and downgradable navigation strategies.

[0112] In a preferred embodiment of the present invention, in step S5, the intelligent multi-source autonomous navigation and guidance task processing is mainly achieved by the processor inside the information processing module M1 pre-setting the reference time calculation unit and allocating different internal task processing function time slots through interrupt operations, thereby realizing multi-source autonomous navigation multi-task parallel processing, including but not limited to 1ms tasks, 5ms tasks and 10ms tasks. The 1ms task is not limited to, but includes, MEMS gyroscope raw angular velocity output and online error calibration task; MEMS accelerometer raw specific force output and online error calibration task; the 5ms task includes, but is not limited to, navigation, positioning, and timing signals output by GNSS receivers, magnetic induction intensity signals output by triaxial magnetometers, barometric altitude signals output by barometers, navigation and positioning signals output by polarization sensors; relative distance signals output by data link rangefinders; and image signals output by monocular / binocular vision cameras, wherein the maximum resolution of monocular vision camera image output is 752×480, with a frame rate of 20FPS, and the maximum resolution of binocular vision camera image output is 4416×1242, with a frame rate of 15FPS; the 10ms task includes, but is not limited to, intelligent multi-source autonomous navigation extended Kalman filter operations.

[0113] In a specific embodiment of this invention, an intelligent multi-source autonomous navigation module for unmanned systems is proposed. The module includes an information processing module unit, a power management unit, a multi-source sensing and measurement unit, a safety monitoring unit, an online storage unit, and an expansion module unit. The information processing module unit consists of a main processor and a crystal oscillator. The power management unit receives an external input voltage of 6V to 36V and converts it into a rated voltage via a DC-DC converter to power the various chips within the intelligent multi-source autonomous navigation system. The multi-source sensing and measurement unit consists of multi-source sensing sensors such as a micro-inertial gyroscope, a micro-inertial accelerometer, a GNSS receiver, a barometer, a magnetometer, a polarization sensor, a data link rangefinder, a monocular / binocular vision camera, and an expansion module. The safety monitoring unit consists of a watchdog monitoring chip. The online storage unit consists of FLASH memory and RAM memory. The expansion module unit reserves a plug-and-play interface for general transmission equipment protocols, used to expand other sensing sensors, processing sensors, and actuator sensors. The information processing module unit communicates and transmits data with the power management unit, multi-source sensing and measurement unit, safety monitoring unit, online storage unit, and expansion module unit via electrical connection. The power management unit provides instrument power signals to the information processing module unit, multi-source sensing and measurement unit, safety monitoring unit, online storage unit, and expansion module unit via electrical connection.

[0114] Figure 5 is a block diagram of the principle of "software-defined navigation" of an intelligent multi-source autonomous navigation system according to a preferred embodiment of the present invention.

[0115] In a preferred embodiment of the present invention, the "software-defined navigation" approach of the intelligent multi-source autonomous navigation system for unmanned systems is based on micro-nano manufacturing processes and integrates technologies such as microelectromechanical systems, micro-optoelectronics, micro-sensing, micro-radio frequency, and micro-integration. Based on an open architecture, it integrates multi-functional components such as high-precision inertial / GNSS / barometric sensing, processors, power conversion and management at the system level microscale. The system uses scenario applications to drive the online switching and autonomous reconstruction of the navigation algorithm library. It expands other multi-source sensing devices through plug-and-play interfaces to form a miniaturized and intelligent multi-source autonomous navigation system suitable for unmanned systems.

[0116] In a preferred embodiment of the present invention, the "software-defined navigation" of the intelligent multi-source autonomous navigation system mainly includes an autonomous navigation physical layer, an autonomous navigation communication protocol layer, an autonomous navigation system management layer, an autonomous navigation algorithm library layer, and an autonomous navigation scenario application layer. The entire multi-source autonomous navigation system adopts a hierarchical open architecture to achieve multi-source detection, information coordination, overall planning, and parallel development.

[0117] In a preferred embodiment of the present invention, the autonomous navigation physical layer serves as the foundation and key to the intelligent multi-source autonomous navigation system. It develops chip-level hardware such as sensors and processors for multi-source navigation systems. The sensors include MEMS gyroscopes, MEMS accelerometers, GNSS receivers, TMR magnetometers, barometers, etc. On the software side, it includes the modularization of hardware-driven layer software for multi-source perception, multi-source fusion, and autonomous decision-making, as well as the standardization of output transmission protocols.

[0118] In a preferred embodiment of the present invention, the autonomous navigation communication protocol layer serves as a "bridge" for interaction between the processor and other functional elements and components. It achieves communication, scheduling, and management of the autonomous navigation physical layer through the standard protocol of the intelligent multi-source autonomous navigation system. Based on the interface types and communication requirements of each functional unit, the research and development of the data bus and communication protocol are completed, especially the unification of the "standard communication protocol," enabling communication between the processor and various functional components, as well as scalable "plug-and-play" functionality.

[0119] In a preferred embodiment of the present invention, the management layer of the autonomous navigation system is located at the bottom layer of the system processor and performs tasks such as task management, task decomposition, resource management, and clock management. Task management refers to the dynamic management and priority management of various concurrent tasks; task allocation is the decomposition of modules for computational processing of a specific task to achieve high-speed parallel computing; resource management is the dynamic scheduling of system storage resources and computing resources according to task characteristics; clock management realizes frequency division and multiplication of the main frequency signal, as well as real-time correction of the time signal; driver management mainly completes the unified management and dynamic configuration of the underlying drivers of different interfaces; and time slot allocation is the internal time slot allocation of embedded execution functions for different tasks and processing cycles.

[0120] In a preferred embodiment of the present invention, the autonomous navigation algorithm library layer mainly completes the research and development of general-purpose algorithms and models, including a model library, a navigation solution algorithm library, an information fusion algorithm library, and a basic code module library. The model library is mainly responsible for modeling interference factors in various complex environments, and modeling the error mechanisms and propagation models of various sensors. The navigation solution algorithm library is mainly responsible for processing and solving information from various sensors and outputting navigation and positioning results, including strapdown inertial navigation systems (SINS), PPP or RTK algorithms for GNSS, MAGCOM algorithms for magnetometers, image processing algorithms for visual cameras, and SLAM algorithms. The information fusion algorithm library is mainly responsible for the fusion and filtering of multi-source information, as well as the estimation and correction of multi-source heterogeneous errors, including a series of Kalman filtering, particle filtering, and factor graph optimization algorithms. The basic code library is mainly responsible for performing general-purpose data reading, data parsing, and standard calculation operations.

[0121] In a preferred embodiment of the present invention, the autonomous navigation scenario application layer is a flexibly configurable application development platform for unmanned system users, specifically including modules such as user-defined tasks, user extended models, user database, user configuration management, health status monitoring, and task execution evaluation. The user-defined task module aims to provide users with a rapid development platform for task description and planning; the user extended model provides users with extensible basic model types and basic configuration interfaces for rapid expansion; the user database provides historical data management, expert knowledge management, etc.; user configuration management handles users' routine parameter configuration and management; health status monitoring is mainly a rapid development platform and tools for monitoring health status for specific tasks; and task evaluation mainly provides standards and criteria for various task evaluations, as well as evaluation algorithms. In this specific embodiment of the present invention, the intelligent multi-source autonomous navigation system for unmanned systems performs online scene switching based on information sources collected by the multi-source perception and measurement module M3 sensors, and autonomously reconstructs the working scene based on the increase or decrease of sensor information sources.

[0122] This invention provides a novel architecture for an intelligent multi-source autonomous navigation system for unmanned systems. It encapsulates an information processing module, power management module, multi-source sensing and measurement module, safety monitoring module, and online storage module within a metal casing. Through a "plug-and-play" open system architecture, it achieves an integrated intelligent multi-source autonomous navigation design. This invention utilizes multi-source sensing sensor data acquisition, including angular rate signals, acceleration signals, barometric altitude signals, magnetic intensity signals, and positioning and timing signals. By fusing and processing multi-source information, it enhances the reliability of the sensing signals and the autonomous navigation capability of the intelligent multi-source autonomous navigation system in complex environments. This invention also provides a high-density integrated intelligent multi-source autonomous navigation system, employing advanced packaging technologies such as TSV, RDL, IPD, and SiP to achieve high-density integration of system modules and three-dimensional system integration, reducing the size, weight, and power consumption of the intelligent multi-source autonomous navigation system. Based on the "software-defined navigation" design concept, this invention fully considers the universality and scalability of the hardware layer, as well as the modularity and integration of the software layer, meeting the future needs of autonomous and controllable technologies for unmanned systems in military-civilian integration.

[0123] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. The appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or their equivalents.

Claims

1. An intelligent multi-source autonomous navigation system, characterized in that... Includes a multi-source sensing and measurement module and an information processing module; The multi-source sensing and measurement module integrates multiple navigation sensors and outputs observations for navigation and positioning. The information processing module receives observations from various navigation sensors, employs corresponding navigation algorithms to obtain the carrier's navigation and positioning information, and denotes a single navigation sensor and its corresponding navigation algorithm as a single navigation information source. Based on the navigation task requirements, the module evaluates the navigation and positioning information output by each single navigation information source to obtain the optimal navigation information source configuration scheme for the current task and scenario. It then selects the optimal information fusion algorithm suitable for the optimal navigation information source configuration from the stored navigation algorithm library and uses the optimal information fusion algorithm to perform multi-source fusion on the output of the optimal navigation information source to obtain optimized navigation and positioning results.

2. The intelligent multi-source autonomous navigation system according to claim 1, characterized in that, The navigation sensors include: an IMU unit, a GNSS receiver, a visual camera, and a magnetometer; The observations of the IMU unit include the angular velocity and acceleration information of the carrier; The observations of the GNSS receiver include pseudorange, carrier phase and differential correction values ​​between the carrier and the navigation satellite. The differential correction values ​​include satellite clock error, receiver clock error, ionospheric delay and tropospheric delay. The observations of a visual camera include environmental image information; The magnetometer's observations include information on the strength and direction of the magnetic field.

3. The intelligent multi-source autonomous navigation system according to claim 2, characterized in that, The information processing module includes a navigation calculation module, an information source configuration optimization module, an availability filtering module, an optimality filtering module, an information fusion module, and a gating switch; The navigation calculation module uses a mechanical arrangement algorithm to obtain the vehicle's velocity, position, and attitude information from the angular velocity and acceleration information output by the IMU unit; it uses a differential positioning algorithm to calculate the vehicle's position and velocity information based on the pseudorange, carrier phase, and differential correction amount output by the GNSS receiver; it uses a SLAM algorithm to calculate the vehicle's position and attitude information based on environmental image information; and it uses a MAGCOM algorithm to calculate the vehicle's position information based on the magnetic field strength and direction information. The information source configuration optimization module acquires navigation and positioning information output from each navigation information source, calculates the quality coefficient of each navigation information source and performs normalization processing. Based on the normalized quality coefficient, which includes accuracy coefficient, real-time coefficient, anti-interference coefficient and fault tolerance coefficient, and combined with the externally input navigation source performance index weights, the optimal navigation information source configuration scheme under the current task scenario is solved through a multi-objective optimization algorithm. The availability filtering module filters out available information fusion algorithms under the optimal navigation information source configuration; The optimality filtering module selects the optimal information fusion algorithm from the available information fusion algorithms; The gating switch sends the output of the configured navigation information source to the information fusion module; The information fusion module, based on the configured navigation information source output, uses the optimal information fusion algorithm to perform multi-source fusion to obtain the fused navigation and positioning results of the carrier.

4. The intelligent multi-source autonomous navigation system according to claim 1, characterized in that, The performance indicators of the navigation source include accuracy, anti-interference ability, real-time performance, and fault tolerance.

5. The intelligent multi-source autonomous navigation system according to claim 1, characterized in that, The anti-interference coefficient is calculated using the following formula: D j =a1η j +a2ε j , In the formula, D j η is the anti-interference coefficient of the j-th navigation information source; a1 and a2 are weighting parameters that can be adjusted based on engineering experience; j ,ε j Let be the system observability and system disturbance degree of the j-th navigation information source, respectively, calculated by the following formula: In the formula, The state observability of the i-th output of the j-th navigation information source; Let r be the state perturbation degree of the i-th output of the j-th navigation information source; r is the number of outputs of the j-th navigation information source, i = 1 to r.

6. The intelligent multi-source autonomous navigation system according to claim 5, characterized in that, State observability It is obtained through the following method: For a single source of navigation information, construct a linear time-invariant system: X k =FX k-1 +w k-1 , Z k =HX k +υ k , In the formula, X k Z represents the state variables of the navigation information source, including the motion state estimation error of the vehicle and related variables of the navigation information source error; F is the state transition matrix; Z represents the state transition matrix. k H is the output of the navigation information source, representing the observed variable; H is the measurement matrix; w k-1 ,υ k These are process noise and measurement noise, respectively. Construct the observability matrix Q and perform SVD singular value decomposition on it: In the formula, Its diagonal elements are the singular values ​​of Q; U is an orthogonal matrix with column vectors QQ. T The eigenvectors of V; V is an orthogonal matrix whose column vectors are Q. T The eigenvectors of Q; Calculate the state observability of the i-th output of the j-th navigation information source: In the formula, σ i Let be the i-th singular value of the observability matrix of the system corresponding to the j-th navigation information source; max(σ i ) represents the maximum value among all its singular values.

7. The intelligent multi-source autonomous navigation system according to claim 5, characterized in that, The state perturbation It is obtained through the following method: For the i-th output of navigation information source j Construct an extended state observer (ESO) and use its output As input to the observer, the ESO extended state observation signal of the i-th output is obtained, the variance of the ESO extended state observation signal of the i-th output is calculated, and the state perturbation degree is calculated: In the formula, Let be the initial observation noise variance of the i-th output of the j-th navigation source. Let be the variance of the ESO extended state observation signal of the i-th output.

8. The intelligent multi-source autonomous navigation system according to claim 1, characterized in that, For the j-th navigation source, the following fault tolerance coefficient is constructed: In the formula, r′ represents the total number of common outputs from all navigation information sources; μ is the i-th common output of the j-th navigation information source; i The common output state z for all navigation sources i The average value.

9. The intelligent multi-source autonomous navigation system for unmanned systems according to claim 1, characterized in that, The optimization problem for finding the optimal navigation sensor configuration is formulated as follows: f1(V)=V·[e 1 ,e 2 …e M ] T f2(V)=V·[T 1 ,T 2 …T M ] T f3(V)=V·[D 1 D 2 …D M ] T f4(V)=V·[Θ 1 ,Θ 2 …Θ M ] T , In the formula, J is the objective function to be optimized; V is a 1×M navigation source configuration vector, where M is the number of navigation sources, and the element in the i-th column of V is 1 to indicate that the i-th navigation source is used, and 0 to indicate that the navigation source is not used; Ω is the optimization decision space, i.e., all possible navigation source configuration vectors; ω1, ω2, ω3, ω4 are the weights of the navigation source performance indicators, representing the priority of navigation source accuracy, real-time performance, anti-interference and fault tolerance in the task requirements, respectively; Here, e represents the normalized navigation source performance function using the min-max method, corresponding to the accuracy, real-time performance, anti-interference, and fault tolerance coefficients, respectively; j T represents the accuracy coefficient of the j-th navigation source, i.e., the average deviation of its common output relative to the mean of the common output of all navigation sources; j Θ is the real-time coefficient of the j-th navigation source, i.e., the time consumed in its information processing; j Let be the fault tolerance coefficient of the j-th navigation source.

10. The intelligent multi-source autonomous navigation system according to claim 1, characterized in that, The specific steps of the availability filtering module are as follows: Construct an availability matching matrix A between navigation solution algorithms and navigation sources. The availability matching matrix is ​​an M×N matrix, where M is the number of navigation sources and N is the number of navigation solution algorithms. The rows of the availability matching matrix represent navigation sources and the columns represent navigation solution algorithms. An element of 1 in the j-th row and i'-th column of the availability matching matrix indicates that the i'-th navigation solution algorithm is applicable to the j-th navigation source, and an element of 0 in the j-th row and j-th column indicates that the i'-th navigation solution algorithm is not applicable to the j-th navigation source. j = 1 to M, i' = 1 to N. Receive the optimal navigation source configuration vector V * The input is a 1×M vector, where the j-th element is 1 to indicate that the j-th navigation source is used, and 0 to indicate that the i'-th navigation source is discarded. Construct a vector of available algorithms B, which is a 1×N vector. The i'th element is 1 to indicate that the i'th information fusion algorithm is available under the current navigation source configuration, and 0 to indicate that it is not available. The navigation source configuration vector and the availability matching matrix are subjected to element-wise AND and XOR operations to obtain the available algorithm vector. The specific calculation method is as follows: B=XNOR(V * ,A&V * ), In the formula, XNOR is the element-wise XOR operation; & is the element-wise AND operation. The output of the availability filtering module is a series of information fusion algorithm vectors decomposed from the availability algorithm vector B, and their relationship conforms to the following formula: B=P1+P2+…P i′ …+P q In the formula, P i′ Let P be a 1×N vector with only one element being 1 and the rest being 0. i′ A value of 1 indicates that it represents the i'th information fusion algorithm; q represents the total number of available information fusion algorithms.

11. The intelligent multi-source autonomous navigation system according to claim 1, characterized in that, The navigation solution algorithm includes extended Kalman filtering, unscented Kalman filtering, volumetric Kalman filtering, federated Kalman filtering, particle filtering, and factor graph optimization.

12. The intelligent multi-source autonomous navigation system according to claim 1, characterized in that, The optimality screening module uses a multi-objective optimization algorithm to obtain the optimal information fusion algorithm: The problem of finding the optimal information fusion algorithm is transformed into a single-objective optimization problem using a weighted method. The specific form of this optimization problem is as follows: f3(P)=P·[ζ 1 ,g 2 …g N ] T , In the formula, J′ is the objective function to be optimized; N is the number of available information fusion algorithms; P is a 1×N vector representing a specific information fusion algorithm, with only one element being 1 and the rest being 0; Ω′ is the decision space, i.e., all available information fusion algorithms; ω′1, ω′2, ω′3 are the weights of the performance indicators of the information fusion algorithms, representing the priority of the information fusion algorithm in terms of accuracy, real-time performance, and anti-interference capability in the task requirements, respectively. These are the quality coefficients of the information fusion algorithm after normalization using the min-max method; is the accuracy coefficient of the j-th algorithm, which is the average offset of its fusion result relative to the fusion results of all algorithms; ζ represents the real-time performance coefficient of the j-th information fusion algorithm, i.e., the time consumed by its computation; j p represents the filtering observability of the j-th information fusion algorithm, i.e., the anti-interference coefficient. i′ This indicates that the i'th information fusion algorithm is available; The covariance matrix Y maintained by the information fusion algorithm can be used as a basis. k To define the filter observability: In the formula Y 0(j′j′) Y represents the element in the j'-th row and j'-th column of the given initial covariance matrix, which is also the standard deviation of the error of the j'-th filter state. k(j′j′) The standard deviation of the filtered state error at time k represents the state at time k. The optimal solution P obtained * The column number i' of the unique non-zero element indicates that the optimal information fusion algorithm is the i'th available information fusion algorithm.

Images