CN-122018307-A - State estimation and bit rate allocation collaborative optimization method of resource-constrained unmanned system

Abstract

A state estimation and bit rate allocation collaborative optimization method of a resource-limited unmanned system includes the steps of firstly establishing a unmanned system dynamics and sensor measurement model, introducing a quantization error model, defining an exponential relation between the model and the bit rate, fusing bounded noise and quantization errors, designing mixed scheduling by static and dynamic segments, transmitting by adopting polling and event triggering protocols, forming a complete measurement sequence by means of a scheduling matrix and a zero-order retainer, designing a rolling time domain estimator, constructing a cost function fused with multiple information to solve optimal estimation, finally constructing a bit rate allocation optimization problem by taking the cost function as a target, and solving by a particle swarm algorithm to realize dynamic allocation under the limitation of the total bit rate. The scheme realizes the state estimation of the unmanned system with high precision, high robustness and real-time under the limitation of strict airborne computing resources and total bit rate.

Inventors

• LI JUNYI
• FENG HAISHEN
• LIU HENG
• XU YONG
• LIU CHANG
• YE YANYAN
• ZHANG BIN
• LU RENQUAN

Assignees

• 广东工业大学

Dates

Publication Date

20260512

Application Date

20251212

Claims (5)

1. 1. The method for collaborative optimization of state estimation and bit rate allocation of the resource-constrained unmanned system is characterized by comprising the following steps: s100, establishing a dynamic model and a sensor measurement model of the unmanned system, wherein the dynamic model and the sensor measurement model comprise a state equation and a measurement equation; describing the quantized measurement value through a component coding and decoding process, introducing a quantization error model, establishing a quantitative relation between a quantization error and an allocated bit rate, determining an exponential relation between a quantization error limit and the bit rate, and finally fusing bounded noise and quantization error augmentation; S200, designing a hybrid scheduling mechanism, dividing sensor nodes into a static section and a dynamic section, wherein the static section adopts a polling protocol to ensure basic data transmission, and the dynamic section adopts a static event triggering protocol with normalized priority to select key nodes for transmission according to the real-time change degree of node data; S300, designing a rolling time domain estimator for quantitative measurement and communication scheduling, constructing a simplified cost function, explicitly fusing quantization errors, scheduling information and historical data, and solving and obtaining an optimal state estimation value through a compensated measurement value and an analysis solving method; S400, constructing an optimization problem with bit rate allocation as a decision variable by taking a cost function of the MHE as an optimization target, solving by adopting a particle swarm optimization algorithm, and realizing dynamic bit rate allocation under the limit of the total bit rate by designing a fitness function and a constraint penalty term.
2. 2. The method for collaborative optimization of state estimation and bit rate allocation for a resource constrained unmanned system according to claim 1, wherein in step S100: Establishing an unmanned system dynamics model for state estimation, referring to a conventional unmanned plane model, wherein a state equation is described as follows: ; wherein the state vector Position in north east coordinate system Attitude angle (roll) Pitching and pitching Yaw ) Linear velocity Sum angular velocity ; Characterizing unmodeled dynamics and external disturbances as process noise; The sensor measurement equation is described as: ; The above matrices A and C are determined by system parameters and are known, and the measurement vectors The sensor is composed of sensor outputs; indicating the measurement noise, the scheme is that And The noise is unknown but bounded, i.e., UBB satisfies: ; ; for the codec we use a component-based codec process, CBCDP, the post-quantization codec metrics can be expressed as: ; Wherein, the Representing the total number of nodes, quantization error for the nodes And an overall quantization error vector Can be expressed as follows: ; ; the further resulting error bound is represented as follows: ; ; Wherein, the Representing a given scaling parameter, is the first The absolute value of the maximum range of each sensor, the specific value being combined with each sensor design, Represent the first Bit rate of individual nodes; the quantization error bound above can be seen as an exponential relationship with the allocated bit rate: ; wherein communication resources are provided And estimating a direct impact of performance A direct quantitative relationship between them, based on which an optimization problem is constructed to systematically allocate bit rate resources; after quantization, quantization errors will be introduced To simplify the subsequent derivation and follow the principle of focusing all bounded perturbations into a single constraint uncertainty vector, one will , And Broadening into a vector And applying a joint norm boundary, Representing nodes Is used for the number of dimensions of (c), Representing the total dimension of the measurement; The post-augmentation redefined system model and the quantized and decoded metrology dynamic expression may be obtained as follows: ; Wherein, the ; ; 。
3. 3. The method for collaborative optimization of state estimation and bit rate allocation for a resource constrained unmanned system according to claim 2, wherein step S200 comprises the steps of: s210, a communication protocol foundation of a hybrid scheduling protocol; s211, modifying a native FlexRay protocol, reserving communication periodic structures of a static section and a dynamic section, and ignoring a symbol window and network idle time; S212, protocol segmentation and node grouping: dividing the FlexRay communication period into a static segment and a dynamic segment, and corresponding to two groups of sensor nodes: static segment correspondence set Scheduling by adopting a polling protocol, namely RRP; dynamic segment correspondence set Scheduling by adopting a redesigned static event triggering protocol, namely SETP; S220, a FlexRay protocol scheduling mechanism; s221, initializing, namely, scheduling matrix Initializing to be an all-zero matrix, wherein the all nodes are in an unscheduled state initially; S222, static segment scheduling, namely RRP, traversing static segment node group Transmission node for determining current time k according to RRP , ; And allocating an identity matrix block to the scheduling matrix, wherein node measurement information which can be successfully transmitted by the static node is expressed as follows by a formula: ; S223, dynamic segment scheduling, namely SETP, traversing dynamic segment node group Initializing a set of dynamic segment nodes that obtain transmission rights Calculating the competition error of each node : ; Wherein, the Representing the successfully decoded and transmitted metric from node j prior to time k, Node normalized priority: ; Wherein, the As a weight matrix, optionally given weight values, The characteristic range is designed according to actual requirements, and can be an expected measurement value, a sensor range and the like; Nodes that exceed the normalized threshold are selected to grant transmission rights, ; Simultaneously updating a set of dynamic segment nodes that obtain transmission rights , ; And allocates an identity matrix block to it in the scheduling matrix, ; The node measurement information that the dynamic node can successfully transmit is expressed as formula (15) through a formula ; S224, combining the scheduling matrix, namely combining the scheduling results of the static segment and the dynamic segment into a final block diagonal scheduling matrix ; ; The measurement information expression of the node successfully transmitted at this time can be expressed as: ; s230, specific operation of a scheduling matrix and a zero-order retainer; The front end of the estimator is used for complementing the missing node measurement value to the last successfully transmitted value of the node by means of a zero-order retainer, namely ZOH, and finally obtaining an estimated end acceptance measurement value The following are provided: ; Wherein, the Is that The data of the whole data segment after the time quantization coding is carried out, and the following measurement expression is received at the front end of the estimator after noise interference is amplified by carrying out the formula (8 c) into the formula (16): ; in the formula, the data received by the estimator Is a complex function including system state Measuring noise Quantization error Scheduling matrix Historical data 。
4. 4. The collaborative optimization method for state estimation and bit rate allocation of a resource constrained unmanned system according to claim 3, wherein in step S300: The measurement used by the estimation end is a complex function containing scheduling information, quantized measurement values and historical data, so that compared with a cost function in a classical MHE objective function, corresponding modification is needed to be made to accurately estimate the state, and the modified objective function is as follows: ; ; The cost function needs to meet the following constraints: ; ; And a noise constraint formula (3) quantization error constraint formula (6); Wherein, the ; Representing known information in the measurement cost term, The measurement value after the compensation is represented, In order to schedule the matrix to be used, Is that A covariance matrix of known prior estimation errors of the moment, wherein diagonal elements are determined according to physical characteristics and prior knowledge of state variables; Metric matrix representing interference level of each sensor node, inverse matrix thereof As a weight matrix in the cost function, the weight matrix is used for punishing measurement deviation; The structure of (1) should be a block diagonal matrix, corresponding to the grouping of sensor nodes, the smaller the interference boundary, the less uncertainty in the measurement of that node, The larger the corresponding diagonal block element value, the corresponding The smaller the diagonal elements of the node measurement channel, thereby giving the node measurement channel a lower penalty weight in the cost function, in particular Nodes of The design is as follows, ; Wherein, the Is the first The upper limit of the measuring noise boundary of the node, the denominator and the other term are the quantization error square form ; The weighting matrix fuses the influence of the measurement noise and the quantization error, so that the estimator can adaptively adjust the trust degree of the data of each node; The final overall interference penalty weight metric matrix may be expressed as ; In equation (18), the objective function compared to MHE lacks a process cost term because of the prior of the present scheme The smooth update mechanism used is The state evolution is implicitly constrained by taking the optimal estimate at the previous moment as a priori; In UBB cases, the goal of MHE is not to minimize the norm of the process noise, but rather to guarantee the final criticality of the estimation error through a thresholding analysis; Unlike recursive filtering methods, which require state augmentation to explicitly model delays caused by communication, rolling-domain estimation naturally adapts to these effects by using fixed-length measurement windows, at each time k, MHE uses received measurement sequences Scheduling information The structural delay in the static section and the zero-order hold behavior in the dynamic section are inherently captured in the metrology sequence; Effectively extracting "new" information from each measurement, while historical values The method avoids the need of state augmentation and fully considers the scheduling effect of the FlexRay protocol; The patent uses the analytical solution of the MHE problem, and can be solved to obtain the MHE problem The optimal estimate of time is as follows: ; the analytical solution is a solution result of the optimization problem of the formula (18), avoids a complex numerical optimization process, and is suitable for real-time operation under the condition of limited airborne calculation of an unmanned system, wherein , ; The calculation flow of the state estimation module can be summarized as each moment The estimator uses data within a time window of fixed length N +1 to derive a target value from the optimization problem by solving the optimization problem To the point of The optimal state estimation sequence of the time instant.
5. 5. The collaborative optimization method for state estimation and bit rate allocation of a resource constrained unmanned system according to claim 3, wherein in step S400: S410, constructing an optimization problem; To realize the optimal allocation of bit rate, a MHE-PSO collaborative optimization algorithm is provided, and the optimization problem is as follows: ; ; , ; Wherein, the Is an objective function (18) of the MHE part, A penalty term for the overall bit rate constraint, Is a sufficiently large penalty factor, selected to be several orders of magnitude greater than a typical MHE cost value, e.g An objective function that ensures that any violation of the total bitrate constraint will have a severe penalty; ensuring that there are enough bits for data encoding after considering the data ID; The optimization problem is solved by the following steps; the fitness function of each particle is defined as: ; Wherein, the , Is made of particles The MHE cost function of the bit rate allocation assessment represented; The function is a specific implementation of an optimization target of a formula (24), the formula (24) directly minimizes the MHE objective function, the MHE objective function is mathematically used as a proxy index of a state estimation error, the minimization of the objective function also indicates the reduction of the state estimation error, and a bit rate allocation scheme with smaller estimation error can be finally obtained after PSO (phase shift keying) is carried out; s420, MHE-PSO collaborative optimization algorithm; s421, initializing, namely randomly initializing the particle position Sum speed of Initializing a personal best location And individual best fitness Global optimum position And global optimum fitness ; S422, fitness evaluation, for each particle i, using an objective function Calculating fitness and applying punishment to the situation of violating constraint; ; In the formula, Corresponding to Penalty term Corresponding to ; S423 updating individual and global optima, updating individual best position of each particle according to the adaptation value And global optimum position ; For the particles obtained in S422 Corresponding fitness function Updating; 1) Updating individual optima if the fitness function of the current particle i Less than the optimal fitness of the individual Then the person is optimally located Assigning a position as current particle i And then the individual is optimally adapted Fitness assigned to the current particle i The formula is expressed as follows: If it is Then ; 2) Updating global optimum, namely if the fitness of the current particle i is smaller than the global optimum fitness, assigning the global optimum position as the position of the current particle i and then assigning the global optimum fitness as the fitness of the current particle i, wherein the formula is as follows: If it is Then ; S424, speed and position updating, adjusting the particle speed using inertial, cognitive and social components, and then updating the particle speed and position; 1) Speed update using inertia Cognitive and social components Generated random numbers To adjust particle velocity The formula is expressed as follows: ; 2) Position update by the particle velocity obtained above The updated particle location formula is expressed as follows: ; s425, forcing constraint of ensuring bit rate to be integer value through proper rounding and clamping operation And bit rate greater than minimum ; S426, terminating, firstly updating the inertia weight in each iteration process From the slave Linearly decrease to The expression is as follows: ; Finally, repeating steps S422-S425 until the maximum number of iterations is reached After iteration is completed A global optimal position is obtained after the next time This is Bit rate allocation scheme for a location as an optimal bit rate allocation scheme 。

Description

State estimation and bit rate allocation collaborative optimization method of resource-constrained unmanned system Technical Field The invention relates to unmanned system control technology, in particular to a state estimation and bit rate allocation collaborative optimization method of a resource-limited unmanned system. Background When an unmanned system performs a task in a complex unknown environment, accurate estimation of various states of the unmanned system is required to ensure the safety of the unmanned system and the completion of the task. However, due to the particular challenges presented by complex unknown environments, state estimation of unmanned systems faces a series of technical difficulties, mainly including the following: The existing unmanned system mostly adopts fixed bit rate allocation or simple polling protocol (RRP), and the static resource allocation mode cannot be adaptively adjusted according to the dynamically changing task scene, environment and real-time perceived value of each sensor node (for example, certain sensor data suddenly becomes abnormally important) of the unmanned system. The direct consequence is that high value data cannot be transmitted in time due to insufficient resources, while low value data continuously occupy valuable communication bandwidth. While some Event Triggered Protocols (ETPs) can provide some flexibility, they suffer from deficiencies in ensuring system stability. An unmanned system needs a scheduling mechanism which can ensure real-time performance and has flexibility. In digital communications, quantization errors are unavoidable. The existing method regards quantization errors as fixed noise, and cannot establish a quantitative mathematical relationship between allocation of communication resources (such as bit rates) and final state estimation accuracy. The design concept of quantization error and communication resource splitting results in the loss of the direction of serving the final estimation objective for optimal allocation of communication resources. The existing estimation algorithm can theoretically provide high-precision estimation for coping with uncertainty caused by quantization errors and communication scheduling, and some advanced estimation algorithms (such as optimization algorithms considering complex constraint) can not meet the real-time requirement of state estimation because the calculation complexity of the existing estimation algorithm is far beyond the bearing capacity of the limited airborne calculation resources of an unmanned system. On the contrary, some simple estimation algorithms which have small calculated amount and can run in real time have serious degradation of estimation performance and insufficient precision when dealing with quantization errors and intermittent data packet loss. This has made the prior art either computationally motionless or poorly accurate. Thus, there is a need for an integrated solution that enables deep fusion of communication scheduling, state estimation and bit rate allocation. Disclosure of Invention Aiming at the defects, the invention aims to provide a collaborative optimization method for state estimation and bit rate allocation of a resource-limited unmanned system, so that the unmanned system can realize high-precision, high-robustness and real-time state estimation under the condition of strict airborne computing resource and total bit rate limitation. To achieve the purpose, the invention adopts the following technical scheme: The method for optimizing the state estimation and bit rate allocation of the resource-constrained unmanned system comprises the following steps: s100, establishing a dynamic model and a sensor measurement model of the unmanned system, wherein the dynamic model and the sensor measurement model comprise a state equation and a measurement equation; describing the quantized measurement value through a component coding and decoding process, introducing a quantization error model, establishing a quantitative relation between a quantization error and an allocated bit rate, determining an exponential relation between a quantization error limit and the bit rate, and finally fusing bounded noise and quantization error augmentation; S200, designing a hybrid scheduling mechanism, dividing sensor nodes into a static section and a dynamic section, wherein the static section adopts a polling protocol to ensure basic data transmission, and the dynamic section adopts a static event triggering protocol with normalized priority to select key nodes for transmission according to the real-time change degree of node data; S300, designing a rolling time domain estimator for quantitative measurement and communication scheduling, constructing a simplified cost function, explicitly fusing quantization errors, scheduling information and historical data, and solving and obtaining an optimal state estimation value through a compensated measurement value and an analysis sol