Bucket wheel machine multi-node cooperative control system and method based on wireless ad hoc network

Abstract

The application discloses a bucket wheel machine multi-node cooperative control system and method based on wireless ad hoc network, which first utilizes multi-dimensional sensor data to solve the local high-order motion state, generates the local desired trajectory sequence containing future intention in the rolling prediction time domain and broadcasts it. The receiving end analyzes the adjacent node trajectory, combines the physical layer signal characteristics and data time effectiveness index to calculate the link reliability coefficient, performs the spatial geometric inflation processing based on uncertainty on the neighbor trajectory, and constructs the safety envelope dynamically matched with the current communication quality. On this basis, the nonlinear model prediction control solver is used to solve the local trajectory with the weighted neighbor trajectory set as the dynamic hard constraint, so as to output the execution action instruction capable of avoiding collision and maximizing operation continuity.

Description

Technical Field

[0001] This application relates to the fields of industrial automation control and industrial Internet of Things (IIoT) technology, specifically to a multi-node collaborative control system and method for bucket wheel excavators based on a wireless ad hoc network. Background Technology

[0002] With the advancement of digital transformation in large bulk cargo ports and power plant storage yards, bucket wheel stacker-reclaimers, as core bulk material conveying equipment, directly determine the turnover efficiency and storage density of storage yards through their automated multi-machine collaborative operation. To overcome the physical limitations of traditional fiber optic reels or sliding contact lines on equipment movement range and improve the flexibility of cluster deployment, industrial wireless ad hoc networking technology is gradually replacing wired communication, becoming the mainstream communication method for mobile bucket wheel stacker clusters to achieve location sharing, task scheduling, and collaborative collision avoidance.

[0003] However, in actual operating environments, the massive metal truss structure and cantilever rotation of the bucket wheel excavator can cause severe interference to the wireless channel through multipath effects and signal shielding, resulting in highly dynamic changes in network topology and extremely unstable link quality. Most existing multi-machine cooperative control technologies are based on ideal communication assumptions, employing a rigid anti-collision logic of real-time position broadcasting plus static distance determination. That is, each node simply calculates the Euclidean distance based on the received GNSS coordinates of its neighbors. This mechanism heavily relies on the determinism of data transmission and struggles to address the asynchronous position data caused by network latency jitter. When the wireless network experiences momentary congestion or continuous packet loss, the neighbor positions obtained by the receiver often lag behind their actual physical state, resulting in significant spatiotemporal deviations. To ensure safety when data is unreliable, existing systems are forced to adopt extremely conservative strategies, pre-setting huge static safety envelopes or employing circuit-breaking protection mechanisms that require immediate stop upon signal interruption. This control method, which lacks the ability to sense and adaptively adjust the link status, not only wastes valuable yard space resources due to excessive safety redundancy distance, but also causes equipment to frequently trigger false alarms or emergency braking in weak network environments, disrupting the continuity of operations and accelerating mechanical wear, making it difficult to meet the actual needs of high-density, high-dynamic collaborative operations. Summary of the Invention

[0004] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a multi-node collaborative control system and method for bucket wheel excavators based on a wireless ad hoc network.

[0005] According to one aspect of this application, a multi-node cooperative control method for a bucket wheel excavator based on a wireless ad hoc network is provided, comprising: Step S1: Perform time alignment and coordinate mapping calculation on the acquired multi-dimensional sensor data stream of the bucket wheel excavator to obtain the current state vector of the excavator containing real-time position, velocity and acceleration information; Step S2: Based on the preset task configuration, perform forward planning on the current state vector of the machine within the set prediction time domain to generate the expected trajectory sequence of the machine containing multiple future discrete time steps; Step S3: While parsing the compressed trajectory data broadcast by the received neighboring nodes to obtain the neighboring node trajectory sequence, calculate the link reliability coefficient based on the physical layer signal characteristics, and use the link reliability coefficient to perform channel quality weighting on the neighboring node trajectory sequence to obtain a weighted neighboring trajectory set. Step S4: Perform spatiotemporal conflict detection and constraint optimization on the local desired trajectory sequence and the weighted neighbor trajectory set to obtain the optimal safe trajectory sequence; Step S5: Perform closed-loop deviation feedback calculation on the optimal safe trajectory sequence based on the current state vector of the machine to obtain the execution commands for driving the bucket wheel excavator trolley travel and cantilever slewing mechanism.

[0006] According to another aspect of this application, a multi-node collaborative control system for a bucket wheel excavator based on a wireless ad hoc network is provided, comprising: The time alignment and coordinate mapping solution module is used to perform time alignment and coordinate mapping solution on the acquired multi-dimensional sensor data stream of the bucket wheel excavator to obtain the current state vector of the machine containing real-time position, velocity and acceleration information; The forward planning module is used to perform forward planning on the current state vector of the machine within a set prediction time domain according to the preset task configuration, so as to generate the expected trajectory sequence of the machine containing multiple discrete time steps in the future. The weighted neighbor trajectory set acquisition module is used to parse the compressed trajectory data broadcast by the received neighbor nodes to obtain the neighbor node trajectory sequence, calculate the link reliability coefficient based on the physical layer signal characteristics, and use the link reliability coefficient to perform channel quality weighting on the neighbor node trajectory sequence to obtain the weighted neighbor trajectory set. The spatiotemporal conflict detection and constraint optimization solution module is used to perform spatiotemporal conflict detection and constraint optimization on the local desired trajectory sequence and the weighted neighbor trajectory set to obtain the optimal safe trajectory sequence. The closed-loop deviation feedback calculation module is used to perform closed-loop deviation feedback calculation on the optimal safe trajectory sequence based on the current state vector of the machine, so as to obtain the execution action commands for driving the bucket wheel excavator trolley travel and cantilever slewing mechanism.

[0007] Compared with existing technologies, this invention proposes a multi-node collaborative control system and method for bucket wheel excavators based on a wireless ad hoc network. First, it utilizes time alignment and kinematic calculations of sensor data to construct the full-dimensional state of the excavator, and generates a desired trajectory sequence containing future motion intentions based on rolling time-domain planning. During cross-node interaction, not only is trajectory data transmitted, but physical layer signal characteristics and data timeliness indicators are also introduced to quantitatively evaluate the reliability of the communication link. Based on the evaluation results, the system performs uncertainty-based spatial covariance dilation processing on the received trajectories of neighboring nodes, mapping the instability of network transmission in real time into a dynamically adjusted geometric safety envelope. Using this envelope as the hard constraint boundary of nonlinear programming, the solver automatically optimizes and generates the optimal control command that balances obstacle avoidance safety and operational continuity in the prediction time domain, achieving a flexible control effect that smoothly maintains equipment operation by adaptively expanding the safety distance under weak network or packet loss conditions. Attached Figure Description

[0008] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 This is a flowchart of a multi-node cooperative control method for a bucket wheel excavator based on a wireless ad hoc network, according to an embodiment of this application. Figure 2 This is a schematic diagram of data flow in the multi-node cooperative control method for bucket wheel excavators based on a wireless ad hoc network according to an embodiment of this application; Figure 3 This is a flowchart illustrating the process of generating a desired trajectory sequence of a bucket wheel excavator based on a wireless ad hoc network multi-node cooperative control method according to an embodiment of this application. The method involves performing forward planning on the current state vector of the excavator within a set prediction time domain based on a preset task configuration. Figure 4 This document describes a flowchart illustrating the process of parsing compressed trajectory data broadcast by neighboring nodes to obtain a neighboring node trajectory sequence, calculating a link reliability coefficient based on physical layer signal characteristics, and using the link reliability coefficient to perform channel quality weighting on the neighboring node trajectory sequence to obtain a weighted neighboring trajectory set, according to an embodiment of this application. Figure 5This is a flowchart illustrating the spatiotemporal conflict detection and constraint optimization solution of the desired trajectory sequence and weighted neighbor trajectory set of the bucket wheel excavator based on a wireless ad hoc network, according to an embodiment of this application, to obtain the optimal safe trajectory sequence. Figure 6 This is a block diagram of a multi-node collaborative control system for a bucket wheel excavator based on a wireless ad hoc network, according to an embodiment of this application. Detailed Implementation

[0010] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0011] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0012] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0013] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0014] Existing bucket wheel excavator collaborative control systems suffer from highly unstable link quality and significant latency jitter when using wireless ad hoc networks for multi-machine interaction. This is due to multipath effects and signal shielding caused by large metal structures at the work site. This communication uncertainty makes traditional anti-collision logic based on instantaneous position broadcasting and static safety distance determination prone to spatiotemporal deviations. Furthermore, it forces the system to pre-set excessively large avoidance buffers or frequently trigger emergency stops upon packet loss to ensure safety, severely restricting the efficiency of high-density continuous operations in the storage yard. Therefore, this application proposes a multi-node collaborative control method for bucket wheel excavators based on wireless ad hoc networks. This method aims to transform the instability of network transmission into quantifiable control constraints, thereby achieving flexible and safe operation in weak network environments. Specifically, it first constructs the full-dimensional state of the machine using kinematic calculations and generates a desired trajectory sequence containing future motion intentions based on rolling time-domain planning. In the cross-node collaboration stage, it innovatively introduces a link reliability assessment mechanism. The receiver calculates the dynamic link reliability coefficient by analyzing the physical layer signal characteristics and data timeliness indicators, and performs spatial covariance expansion processing based on uncertainty on the predicted trajectory of neighboring nodes accordingly. That is, the worse the network quality, the larger the artificially constructed virtual safety envelope. Finally, the nonlinear model predictive controller uses this dynamically changing safety envelope as a hard constraint boundary to automatically find the optimal control command that balances obstacle avoidance safety and operation continuity in the prediction time domain. Thus, under the premise of ensuring physical collision avoidance safety, it effectively avoids unnecessary downtime caused by communication fluctuations by adaptively adjusting the safety distance, thereby improving the robustness and efficiency of the cluster collaborative operation.

[0015] Figure 1 This is a flowchart of a multi-node collaborative control method for a bucket wheel excavator based on a wireless ad hoc network, according to an embodiment of this application. Figure 2 This is a schematic diagram of data flow in a multi-node cooperative control method for a bucket wheel excavator based on a wireless ad hoc network, according to an embodiment of this application. Figure 1 and Figure 2As shown, the multi-node cooperative control method for a bucket wheel excavator based on a wireless ad hoc network according to an embodiment of this application includes the following steps: S1, performing time alignment and coordinate mapping calculation on the acquired multi-dimensional sensor data stream of the bucket wheel excavator to obtain the current state vector of the machine containing real-time position, speed, and acceleration information; S2, performing forward planning on the current state vector of the machine within a set prediction time domain according to a preset task configuration to generate a desired trajectory sequence of the machine containing multiple future discrete time steps; S3, parsing the compressed trajectory data broadcast by the received neighboring nodes to obtain the trajectory sequence of the neighboring nodes, while calculating the link reliability coefficient based on the physical layer signal characteristics, and using the link reliability coefficient to perform channel quality weighting on the trajectory sequence of the neighboring nodes to obtain a weighted neighbor trajectory set; S4, performing spatiotemporal conflict detection and constraint optimization on the desired trajectory sequence of the machine and the weighted neighbor trajectory set to obtain the optimal safe trajectory sequence; S5, performing closed-loop deviation feedback calculation on the optimal safe trajectory sequence based on the current state vector of the machine to obtain the execution action commands for driving the trolley travel and cantilever slewing mechanism of the bucket wheel excavator.

[0016] Specifically, in step S1, the acquired multi-dimensional sensor data stream of the bucket wheel excavator is time-aligned and coordinate-mapped to obtain the current state vector of the excavator, which includes real-time position, velocity, and acceleration information. It should be noted that because the sensing devices deployed at the bucket wheel excavator's operating site, such as the global navigation satellite system, absolute encoder, and travel frequency converter, operate independently, the acquired multi-dimensional sensor data stream suffers from sampling frequency differences and signal transmission delays. Furthermore, the original physical signals inevitably carry mechanical vibration noise and measurement errors. Based on this, the technical solution of this application first performs time alignment and coordinate mapping on the acquired multi-dimensional sensor data stream of the bucket wheel excavator to eliminate time phase deviations caused by hardware asynchrony and to uniformly map heterogeneous physical quantities into a standardized kinematic state space. Through the above processing, a high-precision, real-time, full-dimensional state reference for the excavator can be effectively constructed, providing reliable input data for subsequent trajectory planning and conflict detection, and avoiding control oscillations or calculation errors caused by signal noise and time lag.

[0017] More specifically, in a specific example of this application, step S1 includes: performing timestamp alignment processing and noise reduction cleaning operations based on linear interpolation on the acquired multidimensional sensor data stream to obtain synchronously cleaned sensor data; performing spatial mapping calculations including coordinate translation and rotation transformations on the synchronously cleaned sensor data to obtain bucket wheel end pose data; using a Kalman filter to estimate the acceleration of the velocity components in the bucket wheel end pose data, and assembling the calculated global position coordinates, filtered linear velocity, estimated acceleration, and attitude angle data into vectors according to a preset communication protocol order to obtain the current state vector of the machine.

[0018] In other words, more specifically, the data processing flow strictly follows the logical sequence from time-domain synchronization to spatial-domain resolution and then to state enhancement. First, the processor receives raw multi-dimensional sensor data streams containing the trolley's reference coordinates, cantilever rotation angle, pitch angle, and travel speed. Using the system time of the current control cycle as a reference, it performs timestamp alignment processing based on linear interpolation on data sources with different sampling frequencies. For low-frequency satellite positioning data, interpolation calculations are performed using sampling points from adjacent times to match the control cycle. Simultaneously, a second-order Butterworth low-pass filter is used to perform noise reduction and cleaning operations on the aligned data, removing high-frequency glitches caused by machine vibration, thus obtaining synchronized and cleaned sensor data with a signal-to-noise ratio that meets control requirements. Subsequently, based on the forward kinematics model of the bucket wheel excavator, spatial mapping calculations involving coordinate translation and rotation transformations are performed on the synchronized and cleaned sensor data. This process uses the trolley's reference coordinates as the origin of the local coordinate system. A rotation matrix is ​​used to map the cantilever length vector from joint space to Cartesian space, and base height compensation is superimposed to accurately calculate the absolute position and attitude of the bucket wheel cantilever end's center of gravity in the global coordinate system, obtaining the bucket wheel end's pose data. Finally, to meet the requirements of model predictive control for the system's dynamic characteristics, a Kalman filter is used to estimate the acceleration of the velocity components in the bucket wheel end's pose data. An instantaneous acceleration value is optimally estimated from noisy velocity measurements using a state observer. The calculated global position coordinates, filtered linear velocity, estimated acceleration, and current attitude angle data are then vectorized according to a preset communication protocol to generate a standardized local current state vector, completing the transformation from the original signal to the system state space.

[0019] Specifically, in step S2, based on the preset task configuration, forward planning is performed on the current state vector of the machine within the set prediction time domain to generate a sequence of expected trajectories for the machine containing multiple discrete time steps in the future. It should be noted that, given that bucket wheel stacker-reclaimers are typical high-inertia special-purpose equipment, their start-stop and rotation actions have significant physical lag. Furthermore, the multi-hop transmission of wireless ad hoc networks in industrial settings introduces a non-negligible time delay. Relying solely on the current position status at a single moment for collaborative judgment is prone to inadequate obstacle avoidance response or misjudgment due to outdated information. Therefore, the technical solution of this application further analyzes the target stacker-reclaim position based on the preset task configuration, using the current state vector of the machine as the starting point of evolution, and performs forward planning that satisfies mechanical dynamic constraints within the set prediction time domain. This generates a sequence of expected trajectories for the machine containing multiple discrete time steps in the future, thereby expanding the description of the equipment state from a static current coordinate dimension to a dynamic future intention dimension, broadcasting the spatiotemporal trajectory to be occupied to other nodes in the network instead of just the instantaneous position. Through the above processing, the time delay introduced by the communication link can be effectively compensated, enabling the cooperative control logic to have predictive capabilities and adjust the motion strategy in advance before a potential collision occurs, thus avoiding mechanical impact and operation interruption caused by emergency braking.

[0020] Figure 3 This is a flowchart illustrating the process of generating a desired trajectory sequence for a bucket wheel excavator based on a wireless ad hoc network, according to a preset task configuration, within a set prediction time domain, by performing forward planning on the current state vector of the excavator. The process involves planning based on a preset task configuration. For example,... Figure 3 As shown, step S2 includes: S21, performing path planning and time-domain discretization sampling on the current state vector of the machine and the target position resolved in the task configuration to satisfy preset geometric boundary conditions to obtain a discrete reference path point set; S22, using the discrete reference path point set as the tracking target, performing finite-domain optimization calculation on it using a quadratic programming solver to obtain the nominal control input sequence; S23, using the current state vector of the machine as the evolution starting point, performing forward state recursion calculation on the nominal control input sequence to generate the desired trajectory sequence of the machine.

[0021] In step S21, path planning and time-domain discretization sampling that satisfy preset geometric boundary conditions are performed on the target position resolved from the current state vector of the machine and the task configuration to obtain a discrete reference path point set. It should be noted that since the task configuration executed by the bucket wheel stacker-reclaimer only provides discrete endpoint stacking / reclaiming coordinates and lacks a description of the motion constraints in the intermediate process, direct point-to-point control can easily lead to abrupt changes in the mechanism's motion or exceed the geometric operating boundary of the stockpile. Based on this, the technical solution of this application further performs path planning and time-domain discretization sampling that satisfy preset geometric boundary conditions on the target position resolved from the current state vector of the machine and the task configuration to construct an ideal guiding path that meets both the requirements of mechanical kinematic smoothness and is within the safe operating space of the stockpile. Through the above processing, a discrete reference benchmark with temporal attributes can be effectively provided for subsequent model predictive control, ensuring that the generated control commands always tend towards the operational target at the macroscopic level.

[0022] More specifically, in a concrete example of this application, the processing logic first parses the target 3D coordinates for the next work cycle from the job task configuration issued by the host computer and sets them as the terminal constraint condition for path planning; simultaneously, it uses the real-time position and velocity in the current state vector of the machine as the initial constraint condition. Subsequently, a continuous geometric curve is fitted between the initial point and the target point using a fifth-order polynomial interpolation algorithm. In this process, preset geometric boundary conditions are introduced, which cover the maximum rotation angle of the cantilever, the pitch range, and the spatial constraints of the stacking operation surface, ensuring that the generated curve lies within the feasible region throughout. Finally, based on the prediction time domain length and control step size set by the model predictive control, time-domain discretization sampling is performed on the continuous geometric curve, and a set of discrete reference path points containing position, desired velocity, and attitude information is extracted according to the time series, which serves as the input to the tracking objective function of the subsequent optimization solver.

[0023] In step S22, using the discrete reference path point set as the tracking target, a quadratic programming solver is used to perform finite-time domain optimization calculations to obtain the nominal control input sequence. It should be noted that, as a heavy industrial piece of equipment, the bucket wheel stacker-reclaimer's motion is limited by its enormous mechanical inertia and motor drive power limits, making it impossible to directly conform to the geometric reference path in a transient response manner. Furthermore, unverified control commands are highly susceptible to causing mechanical overload or dynamic overshoot. Therefore, the technical solution of this application further uses the discrete reference path point set as the tracking target and employs a quadratic programming solver to perform finite-time domain optimization calculations. This allows for the solution of the optimal control variable sequence that minimizes the tracking deviation in the prediction time domain, while satisfying the machine's dynamic constraints. Through the above processing, a nominal control input sequence that conforms to the physical system characteristics can be effectively generated, ensuring the smoothness of equipment operation and reducing mechanical wear, providing a dynamically feasible benchmark for subsequent trajectory prediction.

[0024] More specifically, in a concrete example of this application, the computation process constructs a standard quadratic programming problem based on a model predictive control architecture. First, a quadratic cost function is established, including a state tracking error term and a control gain penalty term. The tracking error term measures the Euclidean distance between the predicted trajectory and the reference path point, while the control penalty term limits drastic changes in the control variables to ensure smooth operation. Simultaneously, the maximum linear velocity of the trolley, the maximum angular velocity of the cantilever rotation, and the maximum allowable acceleration and jerk limits for each mechanism are transformed into linear inequality constraints. Then, the aforementioned cost function and constraints are input into a quadratic programming solver. The solver performs rolling optimization iterations in the prediction time domain within each control cycle, searching for a set of velocity and angular velocity commands that minimize the cost function value, thereby outputting a nominal control input sequence that only considers its own kinematic performance and does not yet include interactive obstacle avoidance logic.

[0025] In step S23, using the current state vector of the local machine as the evolution starting point, a forward state recursion calculation is performed on the nominal control input sequence to generate the local machine's desired trajectory sequence. It should be noted that since the nominal control input sequence calculated in the preceding steps is essentially a set of driving commands in the velocity and angular velocity domains, it lacks direct spatial geometric representation and cannot be directly used by neighboring nodes in a wireless ad hoc network for physical collision avoidance calculations and distance determination. Based on this, the technical solution of this application further uses the current state vector of the local machine as the evolution starting point to perform a forward state recursion calculation on the nominal control input sequence, thereby using the system's discrete state transition equations to map the abstract control variables back to the intuitive Cartesian coordinate space and joint attitude space. Through the above processing, a local machine's desired trajectory sequence containing multiple future discrete time steps can be effectively generated. This sequence not only carries predicted position information that conforms to dynamic constraints but also provides a standardized spatiotemporal data carrier for subsequent network broadcasting and multi-machine spatial conflict detection.

[0026] More specifically, in a concrete example of this application, the process performs a model-based open-loop predictive simulation. The processor first reads the global coordinates and attitude angles from the current state vector as the initial state boundary at time zero, and calls a pre-set discrete kinematic model of the bucket wheel excavator as the core of the state evolution. Subsequently, the recursive logic is expanded step-by-step according to the time step of the prediction time domain. At each discrete time node, the corresponding linear velocity and angular velocity commands in the nominal control input sequence are substituted into the state transition equation, and the predicted state from the previous time step is integrated to calculate the displacement increment of the trolley and the cantilever rotation angle increment for the next time step. Finally, the global position coordinates, attitude orientation, and corresponding velocity vectors obtained at each prediction time step are sequentially encapsulated in time order to form a continuous, spatiotemporally consistent sequence of the machine's desired trajectory, providing a definite geometric description for subsequent data compression broadcasting and collision constraint construction.

[0027] Specifically, in step S3, while parsing the compressed trajectory data broadcast by the received neighboring nodes to obtain the trajectory sequence of neighboring nodes, a link reliability coefficient is calculated based on the physical layer signal characteristics. This link reliability coefficient is then used to weight the neighboring node trajectory sequence by channel quality to obtain a weighted set of neighboring trajectories. It should be noted that, given the large metal truss structures and dynamic rotating mechanisms densely distributed at the bucket wheel excavator operation site, which can cause severe multipath fading and obstruction effects on wireless signals, the quality of the communication link between nodes exhibits high time-varying characteristics. The received neighboring machine trajectory data often carries significant transmission delays or potential packet loss / interruption risks. If rigid obstacle avoidance is performed solely based on the calculated geometric coordinates, collision misjudgments are highly likely due to decreased data confidence. Based on this, the technical solution of this application further analyzes the compressed trajectory data broadcast by the received neighboring nodes to obtain the trajectory sequence of neighboring nodes. Simultaneously, it calculates the link reliability coefficient based on the physical layer signal characteristics and uses the link reliability coefficient to weight the trajectory sequence of neighboring nodes by channel quality to obtain a weighted neighboring trajectory set. This establishes a mapping relationship between the communication domain state and the physical domain security boundary, quantitatively converting indicators such as link quality indication, received signal strength, and data timeliness into uncertainty inflation factors for predicting the location of neighboring devices. Through the above processing, a dynamic security envelope that adaptively expands and contracts with channel quality can be effectively constructed. When the network environment deteriorates or data becomes outdated, the virtual avoidance radius for neighboring devices is automatically expanded, forcing the local control strategy to reserve a larger safety buffer space during the planning stage. This achieves flexible fault-tolerant control for weak network conditions while ensuring physical security.

[0028] Figure 4This document describes a flowchart illustrating the process of parsing compressed trajectory data broadcast by neighboring nodes to obtain a neighboring node trajectory sequence, calculating link reliability coefficients based on physical layer signal characteristics, and then using these link reliability coefficients to apply channel quality weighting to the neighboring node trajectory sequence to obtain a weighted neighboring trajectory set, according to an embodiment of this application. Figure 4 As shown, step S3 includes: S31, performing protocol deframing on the received neighbor node broadcast data stream to obtain a set of neighbor geometric trajectories; S32, generating a timestamp based on the data parsed from the neighbor node broadcast data stream to calculate an information age index, and performing a comprehensive weighted calculation of timeliness and signal quality with the link quality indicator in the normalized physical layer signal characteristics to generate a link reliability score set; S33, using the link reliability score set, performing spatial sphere expansion processing based on dynamic safety radius on each predicted location point in the set of neighbor geometric trajectories to obtain a weighted neighbor trajectory set.

[0029] In step S31, the received neighbor node broadcast data stream is deframed using a protocol to obtain a set of neighbor geometric trajectories. It should be noted that because the data packets transmitted in the wireless ad hoc network undergo protocol encapsulation at the physical and link layers, and to save valuable air interface bandwidth, the trajectory information broadcast by the neighbor is parameterized and compressed into Bezier control points at the sending end. Therefore, the local controller cannot directly use the original binary code stream for intuitive spatial location determination. Based on this, the technical solution of this application further performs protocol deframed processing on the received neighbor node broadcast data stream to obtain a set of neighbor geometric trajectories. This removes the additional overhead of the communication protocol and inversely reconstructs the fine motion path of the neighboring bucket wheel machine in the future time domain from the compressed parameters. Through the above processing, abstract communication data can be effectively restored to a concrete geometric description, ensuring that subsequent anti-collision algorithms can obtain accurate spatial occupancy information of the neighboring machines.

[0030] More specifically, in a concrete example of this application, the data link layer processing unit first captures the raw bit stream demodulated by the radio frequency module and verifies the integrity of the data frame using a cyclic redundancy check (CRC) algorithm, automatically discarding error packets that fail the check. For valid data frames that pass the check, the frame header and trailer are removed according to the ad hoc network communication protocol to extract the application layer payload data. Subsequently, the source node identifier, the absolute timestamp at the time of data generation, and the coordinates of four three-dimensional Bézier control points representing the expected trajectory shape of neighboring machines are parsed from the payload. Next, the processing logic calls a preset third-order Bézier curve generation algorithm, using the above four control points as geometric constraints, to map the normalized time parameters to the trajectory space, and inversely reconstructs a series of discrete coordinate point sequences in the future prediction time domain through interpolation calculation. Finally, the reconstructed set of neighboring geometric trajectories containing the position sequence and corresponding time labels is stored in a dynamic object list, providing standardized geometric input for subsequent weighted calculations combined with link quality.

[0031] In step S32, an information age index is calculated based on the timestamp generated from the data parsed from the neighbor node broadcast data stream. This age index is then weighted and calculated in conjunction with the link quality indicator in the normalized physical layer signal characteristics to generate a link reliability score set. It should be noted that due to the complex electromagnetic environment in industrial settings and the drastic fluctuations in network topology caused by the movement of bucket wheel excavators, received neighbor node data may not only experience signal strength attenuation but also transmission delays due to multi-hop routing queuing. Therefore, the physical layer signal strength alone cannot accurately reflect the validity and real-time performance of service data. Based on this, the technical solution of this application further calculates an information age index based on the timestamp generated from the data parsed from the neighbor node broadcast data stream and weights it in conjunction with the link quality indicator in the normalized physical layer signal characteristics to quantitatively evaluate the confidence of the communication link from two dimensions: data freshness and channel physical health. Through the above processing, a link reliability score set that reflects the current network status can be effectively generated, preventing outdated or high-error-risk location data from misleading collaborative control strategies and ensuring that security decisions are always based on highly reliable information sources.

[0032] More specifically, in a concrete example of this application, the evaluation process employs a multi-index fusion algorithm to construct a scalarized trust score. First, the communication processing unit extracts the absolute generation timestamp of the neighboring node when generating the trajectory data from the received broadcast data frame payload and reads the current absolute system time. The difference between the two is used to obtain the information age index, which directly characterizes the degree of data lag during network transmission. Simultaneously, the physical layer driver reads the link quality indicator value recorded in the register of the wireless RF chip at the moment the data packet is received, compares it with the hardware-defined maximum theoretical value, and normalizes it to obtain a signal quality factor reflecting the current channel bit error rate level. Subsequently, a weighted fusion model based on exponential decay is used to comprehensively calculate the two components, where the signal quality factor serves as the baseline amplitude, and the information age index serves as the time variable for the exponential decay term. This causes the data reliability score to decrease non-linearly with increasing transmission delay and is directly modulated by the instantaneous signal strength. Finally, the calculated values ​​between zero and one are encapsulated as a link reliability score set, serving as a key gain coefficient for subsequent adjustment of the safety envelope radius.

[0033] In step S33, the link reliability score set is used to perform spatial spherical expansion processing based on dynamic safety radius on each predicted location point in the neighbor geometric trajectory set to obtain a weighted neighbor trajectory set. It should be noted that since the data transmitted by the wireless ad hoc network is essentially a discrete sampling observation of the motion state of neighboring bucket wheel machines, and the predicted trajectory points after network transmission have accumulated errors in both the time and spatial dimensions, directly using the calculated geometric coordinates as zero-dimensional absolute positions for obstacle avoidance judgment ignores the uncertainty of position information in weak network environments. This can easily lead to serious mechanical collision accidents due to small deviations between the actual physical position and the predicted point. Therefore, the technical solution of this application further utilizes the link reliability score set to perform spatial spherical expansion processing based on dynamic safety radius on each predicted location point in the neighbor geometric trajectory set. This quantitatively transforms the quality fluctuations of the communication link and the time delay of data transmission into physical defense boundaries in geometric space, i.e., it compensates for the unreliability of information by forcibly expanding the range of position uncertainty through an algorithm. Through the above processing, a set of adaptive safety pipeline constraints can be effectively constructed, enabling the control system to automatically maintain a larger physical interval when the channel quality deteriorates, and allow closer collaborative operation when communication is good, thus achieving a dynamic balance between safety and operational efficiency.

[0034] More specifically, in a concrete example of this application, the processing procedure employs a multi-factor fusion dynamic radius calculation model to geometrically reconstruct each discrete prediction point. The calculation logic first traverses each prediction time step in the neighbor geometric trajectory set, reading the corresponding three-dimensional position coordinates and the current running speed of the neighbor node; simultaneously, it calls the link reliability score set generated in the preceding steps and extracts the corresponding normalized confidence coefficient. Subsequently, based on the uncertainty propagation law, an expansion radius calculation formula is constructed, consisting of three parts: first, a static physical reference radius determined by the outer contour of the bucket wheel excavator's mechanical structure, which constitutes the hard baseline for collision avoidance; second, a dynamic buffer distance calculated based on the velocity vector, used to cover the braking stroke required for emergency braking; and most importantly, a network uncertainty penalty term modulated by the link reliability coefficient is introduced. This term is negatively correlated with the link score, meaning the lower the link score, the larger the penalty term value, and this penalty term increases linearly with the increase of the prediction time step, reflecting the natural decay of long-term prediction reliability. Finally, the processor appends the calculated dynamic safety radius to the corresponding trajectory coordinates, mathematically expanding the originally discrete set of points into a series of spatial spheres distributed along the time axis, forming a weighted neighbor trajectory set, which serves as an insurmountable hard constraint boundary in the subsequent nonlinear programming solver.

[0035] Specifically, in step S4, spatiotemporal conflict detection and constraint optimization are performed on the local desired trajectory sequence and the weighted neighbor trajectory set to obtain the optimal safe trajectory sequence. It should be noted that since the initially generated desired trajectory is mainly based on static task objective planning and does not fully consider dynamic environmental interference, it is highly likely to intrude into the safety envelope range of neighboring nodes, which may expand significantly due to poor link quality, in the future time domain. Simply relying on traditional distance threshold judgments would frequently trigger unnecessary emergency braking or cause operation stoppages. Based on this, the technical solution of this application further performs spatiotemporal conflict detection and constraint optimization on the local desired trajectory sequence and the weighted neighbor trajectory set. This constructs the weighted trajectories of neighboring nodes into an insurmountable set of time-varying inequalities. A nonlinear model predictive control algorithm is then used to iteratively optimize within the feasible solution space, calculating a corrected path that strictly satisfies the dynamic obstacle avoidance boundary while minimizing the deviation from the original task path. Through the above processing, the optimal safety trajectory sequence can be effectively generated, realizing the transformation from passive trigger-based obstacle avoidance to active predictive avoidance. This ensures that even when network fluctuations force the safety distance to be expanded, the equipment can still maintain continuous and stable safe operation through smooth path fine-tuning or speed avoidance.

[0036] Figure 5This is a flowchart illustrating the spatiotemporal conflict detection and constraint optimization solution of the desired trajectory sequence and weighted neighbor trajectory set of the bucket wheel excavator based on a wireless ad hoc network, according to an embodiment of this application, to obtain the optimal safe trajectory sequence. Figure 5 As shown, step S4 includes: S41, based on the dynamic safety radius and predicted position information contained in the weighted neighbor trajectory set, performing Euclidean distance inequality modeling on the spatial geometric relationship between the local machine and neighboring nodes in the prediction time domain to obtain a collision detection constraint set; S42, with minimizing the tracking deviation from the local machine's expected trajectory sequence as the optimization objective, using the collision detection constraint set as the boundary condition of the nonlinear programming problem, inputting iterative calculation of constraint optimization into the nonlinear model predictive control solver to obtain the optimal control input sequence; S43, using the local machine's current state vector as the initial condition, performing forward simulation playback and state data encapsulation on the optimal control input sequence to obtain the optimal safe trajectory sequence.

[0037] In step S41, based on the dynamic safety radius and predicted position information contained in the weighted neighbor trajectory set, Euclidean distance inequality modeling is performed on the spatial geometric relationship between the machine and its neighboring nodes in the prediction time domain to obtain a collision detection constraint set. It should be noted that since the weighted neighbor trajectory set generated in the preceding steps is essentially only discrete geometric data points containing spatiotemporal coordinates and dynamic expansion radii, it does not directly constitute boundary conditions that can be recognized by a mathematical optimization solver. Furthermore, the dynamic approach process of multiple bucket wheel excavators in three-dimensional space requires a precise mathematical model to define the critical states of safety and danger. Therefore, the technical solution of this application further uses the dynamic safety radius and predicted position information contained in the weighted neighbor trajectory set to perform Euclidean distance inequality modeling on the spatial geometric relationship between the machine and its neighboring nodes in the prediction time domain to obtain a collision detection constraint set. This transforms the physical-level collision avoidance requirements into a rigorous set of algebraic inequalities in a nonlinear programming problem, clearly defining the feasible state space of the machine at each predicted time in the future. Through the above processing, standardized mathematical hard constraints can be effectively provided for subsequent constraint optimization solutions, ensuring that the control commands calculated by the solver logically absolutely meet the dynamic safety distance requirements determined by link quality.

[0038] More specifically, in a concrete example of this application, the modeling process is executed in real time within each rolling optimization calculation cycle of the model predictive controller. The processor first iterates through each discrete time step in the prediction time domain, extracting the position variables from the local state prediction equation at that moment as undetermined parameters, and simultaneously reads the predicted coordinates of neighboring nodes at the corresponding moment in the weighted neighbor trajectory set, as well as the dynamic safety radius weighted by channel quality. Subsequently, based on Euclidean geometry principles, a nonlinear function describing the relative distance between the centers of the local and neighboring nodes in three-dimensional space is constructed. This function calculates the L2 norm of the difference between the undetermined position coordinates of the local node and the known predicted coordinates of the neighboring node. Next, an inequality relationship is established between the calculated real-time Euclidean distance and the corresponding dynamic safety radius. The Euclidean distance must be greater than or equal to the dynamic safety radius at that moment, and this relationship is rearranged into a standard form of nonlinear inequality constraint, i.e., the dynamic safety radius minus the Euclidean distance between the two points is less than or equal to 0. Finally, the numerous inequalities generated for all time steps in the prediction time domain and all neighboring nodes within the communication range are instantiated and aggregated one by one to form a high-dimensional collision detection constraint set. Mathematically, this set encloses a collision-free pipeline that dynamically changes with time and network state, serving as the geometric forbidden zone boundary for the subsequent nonlinear programming solver to search for the optimal solution.

[0039] In step S42, with the optimization objective of minimizing the tracking deviation from the desired trajectory sequence, the collision detection constraint set is used as the boundary condition of the nonlinear programming problem and input into the nonlinear model predictive control solver for constraint optimization iterative calculation to obtain the optimal control input sequence. It should be noted that since the desired trajectory planned in the preceding steps is only generated based on static job tasks and does not include physical avoidance logic for dynamic neighbor nodes, simple logical judgment cannot find a balance between job continuity and safety under multidimensional nonlinear constraints. Furthermore, directly executing the original trajectory is highly likely to violate the dynamic safety boundary determined by network channel quality. Therefore, the technical solution of this application further uses minimizing the tracking deviation from the desired trajectory sequence as the optimization objective, and inputs the collision detection constraint set as the boundary condition of the nonlinear programming problem into the nonlinear model predictive control solver for constraint optimization iterative calculation. This mathematically constructs a constrained optimization space, within which a set of control instructions that can strictly satisfy all dynamic hard constraints and maintain the original job rhythm to the greatest extent is automatically searched. The above-mentioned treatment can effectively solve the problems of deadlock or unnecessary emergency stop in multi-machine collaboration, and realize the flexible obstacle avoidance and automatic reset of equipment in confined spaces.

[0040] More specifically, in a concrete example of this application, the calculation process relies on the core architecture of nonlinear model predictive control, and is implemented in a logical closed loop according to objective construction, constraint injection, and iterative solution. First, the controller constructs a multi-objective weighted scalar cost function, which includes two core penalty terms: one is a state tracking penalty term, used to calculate the sum of squared Euclidean distance deviations between the predicted state and the corresponding reference points in the expected trajectory sequence of the machine; the other is a control increment penalty term, used to suppress drastic changes in control input to ensure smooth operation of the mechanism. Simultaneously, the collision detection constraint set generated in the preceding steps is used as the inequality boundary conditions for the nonlinear programming problem, and the physical and mechanical limitations of the bucket wheel excavator itself, including maximum travel speed, maximum angular velocity, and acceleration limits, are superimposed, together forming the geometric enclosure of the feasible solution space. Subsequently, a pre-set sequential quadratic programming or interior-point method numerical solver is invoked, and the aforementioned cost function and the full set of constraints are input into the solution core. The solver performs multiple iterative calculations within each control cycle, searching for the extreme point that minimizes the cost function through sequential quadratic programming iteration and / or interior-point method iteration while satisfying all hard safety constraints. Finally, the solver outputs a set of optimal control input sequences covering the entire prediction time domain. This sequence contains the trolley speed commands and cantilever slewing commands for multiple discrete future moments, serving as the direct driving basis for the physical actuators.

[0041] In step S43, the optimal control input sequence is subjected to forward simulation playback and state data encapsulation using the current state vector of the local machine as the initial condition to obtain the optimal safe trajectory sequence. It should be noted that since the direct output of the nonlinear model predictive control solver is essentially a sequence of control variables such as velocity and angular velocity, lacking a direct spatial position description, this abstract control instruction set cannot be directly used by the host computer for visualized safety monitoring, nor can it be directly used as the initial reference value for the next cycle optimization problem to accelerate calculation. Based on this, the technical solution of this application further uses the current state vector of the local machine as the initial condition to perform forward simulation playback and state data encapsulation of the optimal control input sequence to obtain the optimal safe trajectory sequence. This allows the optimized control instruction set to be remapped into a continuous state trajectory in the Cartesian coordinate system through model deduction. Through the above processing, a visualized optimal safe trajectory sequence can be effectively obtained, verifying the geometric execution effect of the control strategy and providing a high-quality hot-start initial solution for subsequent rolling optimization, thus improving the convergence speed of the algorithm.

[0042] More specifically, in a concrete example of this application, the processing logic executes a strictly time-sequential model derivation process. First, real-time observations, including the absolute coordinates of the trolley and the cantilever attitude angle, are extracted from the current state vector of the machine and set as the zero-time boundary conditions for the playback simulation. Then, the discrete kinematic state equations of the bucket wheel excavator are invoked as transition operators, and the velocity and angular velocity commands from the optimal control input sequence are progressively applied to the current state variables in the predicted time domain order. Integration is then performed to calculate the state vector for the next discrete time step, containing the position coordinates and velocity direction. Finally, the calculated state data for all predicted time steps are extracted, formatted, and encapsulated. A complete optimal safety trajectory sequence is generated by time indexing and stored in system memory for use in the next step of driving the underlying actuators and updating the reference trajectory buffer for the next optimization cycle.

[0043] Specifically, in step S5, a closed-loop deviation feedback calculation is performed on the optimal safe trajectory sequence based on the current state vector of the machine to obtain the execution commands for driving the bucket wheel excavator's trolley travel and cantilever slewing mechanism. It should be noted that since the discrete trajectory points generated by the nonlinear model predictive control solver are essentially top-level motion planning data, rather than voltage or current control signals that the underlying actuators can directly recognize, and the bucket wheel excavator is affected by nonlinear disturbances such as track friction, wind load, and mechanical clearance during actual operation, simply issuing open-loop commands can easily lead to cumulative drift between the actual position and the planned path. Therefore, the technical solution of this application further performs a closed-loop deviation feedback calculation on the optimal safe trajectory sequence based on the current state vector of the machine to construct a real-time position servo closed loop, transforming the macroscopic trajectory tracking task into microscopic motor speed and hydraulic valve opening adjustment. Through the above processing, control errors caused by external environmental interference can be effectively eliminated, ensuring that the physical machinery strictly conforms to the algorithm-planned safe pipeline operation, achieving a precise mapping from digital calculation to physical action.

[0044] More specifically, in a specific example of this application, step S5 includes: extracting the target state point of the next moment from the optimal safe trajectory sequence, and performing deviation vectorization calculation based on the feedforward gain matrix in combination with the current state vector of the machine to generate an instantaneous control deviation vector; performing control law operation on the instantaneous control deviation vector to generate the original motor drive signal; mapping the original motor drive signal to the controller output register address based on the fieldbus protocol, and sending it to the trolley frequency converter and slewing hydraulic station through the physical hardware interface to drive the actuator to move, thereby outputting the execution action command.

[0045] In other words, more specifically, the execution control process follows a rigorous logic from state deviation extraction to control law calculation and then to hardware bus transmission. The processing logic first extracts the target state point for the next moment from the optimal safe trajectory sequence based on the rolling time-domain control principle, using it as the instantaneous tracking target. This target is then differentially calculated with the current state vector of the machine acquired in real-time by the sensors, and combined with a preset feedforward gain matrix to perform deviation vectorization calculation, thereby generating an instantaneous control deviation vector containing the trolley position error component and the cantilever angle error component. Subsequently, this instantaneous control deviation vector is input into the cascaded PID controller of the position and speed loops for proportional, integral, and derivative control law calculations. This transforms the geometric deviation in the position domain into a command value in the speed or torque domain, and after dead-zone compensation and amplitude limiting processing, generates the original motor drive signal that conforms to the physical characteristics of the motor and hydraulic valve. Finally, based on the industrial fieldbus protocol standard, the digitized original motor drive signal is mapped to the output register address of the programmable logic controller, and the instruction data frame is sent to the trolley travel inverter and the hydraulic station of the slewing mechanism at a high frequency through the physical hardware interface, directly driving the variable frequency motor speed regulation and the proportional solenoid valve to act, thereby outputting the final execution action command to drive the mechanical displacement and completing the control closed loop.

[0046] In summary, the multi-node cooperative control method for bucket wheel excavators based on a wireless ad hoc network, according to embodiments of this application, is explained. First, it utilizes time alignment and kinematic calculations of sensor data to construct the full-dimensional state of the machine, and generates a desired trajectory sequence containing future motion intentions based on rolling time-domain planning. During cross-node interaction, not only is trajectory data transmitted, but physical layer signal characteristics and data timeliness indicators are also introduced to quantitatively evaluate the reliability of the communication link. Based on the evaluation results, spatial covariance dilation based on uncertainty is performed on the received trajectories of neighboring nodes, mapping the instability of network transmission in real time into a dynamically adjusted geometric safety envelope. Using this envelope as the hard constraint boundary of nonlinear programming, the solver automatically optimizes and generates the optimal control command that balances obstacle avoidance safety and operational continuity in the prediction time domain, achieving a flexible control effect that smoothly maintains equipment operation by adaptively expanding the safety distance under weak network or packet loss conditions.

[0047] Furthermore, a multi-node collaborative control system for bucket wheel excavators based on a wireless self-organizing network is also provided.

[0048] Figure 6 This is a block diagram of a multi-node cooperative control system for a bucket wheel excavator based on a wireless ad hoc network, according to an embodiment of this application. Figure 6As shown, the multi-node cooperative control system 100 for bucket wheel excavators based on a wireless ad hoc network according to an embodiment of this application includes: a time alignment and coordinate mapping calculation module 110, used to perform time alignment and coordinate mapping calculation on the acquired multi-dimensional sensor data stream of the bucket wheel excavator to obtain the current state vector of the machine containing real-time position, velocity, and acceleration information; a forward planning module 120, used to perform forward planning on the current state vector of the machine within a set prediction time domain according to a preset task configuration to generate a desired trajectory sequence of the machine containing multiple future discrete time steps; and a weighted neighbor trajectory set acquisition module 130, used to compress the received neighbor node broadcasts. While parsing the trajectory data to obtain the trajectory sequence of neighboring nodes, the link reliability coefficient is calculated based on the physical layer signal characteristics, and the channel quality weighting of the trajectory sequence of neighboring nodes is applied using the link reliability coefficient to obtain a weighted neighboring trajectory set; the spatiotemporal conflict detection and constraint optimization solution module 140 is used to perform spatiotemporal conflict detection and constraint optimization solution on the local desired trajectory sequence and the weighted neighboring trajectory set to obtain the optimal safe trajectory sequence; the closed-loop deviation feedback calculation module 150 is used to perform closed-loop deviation feedback calculation on the optimal safe trajectory sequence based on the local current state vector to obtain the execution action command used to drive the bucket wheel excavator trolley travel and cantilever slewing mechanism.

[0049] The time alignment and coordinate mapping solution module includes: The timestamp alignment and denoising cleaning unit is used to perform timestamp alignment and denoising cleaning operations based on linear interpolation on the acquired multidimensional sensor data stream to obtain synchronized and cleaned sensor data. The spatial mapping solution unit is used to perform spatial mapping solution, including coordinate translation and rotation transformation, on the sensor data after synchronous cleaning to obtain the bucket wheel end pose data. The current state vector acquisition unit is used to estimate the acceleration of the velocity component in the bucket wheel end pose data using a Kalman filter, and to assemble the calculated global position coordinates, filtered linear velocity, estimated acceleration and attitude angle data into vectors according to a preset communication protocol order to obtain the current state vector of the machine.

[0050] The module for obtaining the weighted nearest neighbor trajectory set includes: The protocol deframing unit is used to perform protocol deframing on the received neighbor node broadcast data stream to obtain a set of neighbor geometric trajectories; The link reliability score set generation unit is used to generate a timestamp based on the data parsed from the broadcast data stream of neighboring nodes, calculate the information age index, and perform a comprehensive weighted calculation of timeliness and signal quality with the link quality indication in the normalized physical layer signal characteristics to generate a link reliability score set. The spatial sphere expansion processing unit is used to perform spatial sphere expansion processing based on dynamic safety radius on each predicted location point in the neighbor geometric trajectory set using the link reliability score set, so as to obtain a weighted neighbor trajectory set.

[0051] The spatiotemporal conflict detection and constraint optimization solution module includes: The Euclidean distance inequality modeling unit is used to model the spatial geometric relationship between the local machine and its neighboring nodes in the prediction time domain based on the dynamic safety radius and predicted location information contained in the weighted neighbor trajectory set, so as to obtain the collision detection constraint set. The optimal control input sequence acquisition unit is used to minimize the tracking deviation from the local desired trajectory sequence as the optimization objective. It inputs the collision detection constraint set as the boundary condition of the nonlinear programming problem into the nonlinear model predictive control solver for constraint optimization iterative calculation to obtain the optimal control input sequence. The optimal safety trajectory sequence acquisition unit is used to perform forward simulation playback and state data encapsulation on the optimal control input sequence with the current state vector of the local machine as the initial condition to obtain the optimal safety trajectory sequence.

[0052] As described above, the multi-node collaborative control system 100 for bucket wheel excavators based on a wireless ad hoc network according to embodiments of this application can be implemented in various types of computing devices or control units. For example, it can be a ruggedized industrial computer deployed in the electrical room of the bucket wheel excavator, an embedded communication controller installed on the cantilever slewing platform, or a high-performance edge computing unit integrated in the main control PLC cabinet of the equipment. In one possible implementation, the multi-node collaborative control system 100 for bucket wheel excavators based on a wireless ad hoc network according to embodiments of this application can be integrated into the computing device as a software module and / or a hardware module. For example, the multi-node collaborative control system 100 for bucket wheel excavators based on a wireless ad hoc network can be a resident data processing service in the operating system of the computing device. This software module is configured to perform time alignment and coordinate mapping of multi-dimensional sensor data, forward-looking planning of trajectory sequences based on rolling time domain, neighbor trajectory weighted analysis combined with link quality, and constraint optimization solution based on dynamic safety envelope. Alternatively, it can be a dedicated multi-machine collaborative anti-collision control algorithm program developed for the computing device. Of course, the multi-node collaborative control system 100 of the bucket wheel excavator based on the wireless self-organizing network can also be one of the many hardware modules of the computing device or control unit, or be embedded in the field programmable gate array circuit to accelerate the iterative calculation of nonlinear programming problems in parallel, or be an integrated circuit for processing self-organizing network communication protocols for a specific application.

[0053] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, and are not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A multi-node cooperative control method for a bucket wheel excavator based on a wireless ad hoc network, characterized in that, include: Step S1: Perform time alignment and coordinate mapping calculation on the acquired multi-dimensional sensor data stream of the bucket wheel excavator to obtain the current state vector of the excavator containing real-time position, velocity and acceleration information; Step S2: Based on the preset task configuration, perform forward planning on the current state vector of the machine within the set prediction time domain to generate the expected trajectory sequence of the machine containing multiple future discrete time steps; Step S3: While parsing the compressed trajectory data broadcast by the received neighboring nodes to obtain the neighboring node trajectory sequence, calculate the link reliability coefficient based on the physical layer signal characteristics, and use the link reliability coefficient to perform channel quality weighting on the neighboring node trajectory sequence to obtain a weighted neighboring trajectory set. Step S4: Perform spatiotemporal conflict detection and constraint optimization on the local desired trajectory sequence and the weighted neighbor trajectory set to obtain the optimal safe trajectory sequence; Step S5: Perform closed-loop deviation feedback calculation on the optimal safe trajectory sequence based on the current state vector of the machine to obtain the execution commands for driving the bucket wheel excavator trolley travel and cantilever slewing mechanism.

2. The multi-node cooperative control method for bucket wheel excavators based on a wireless ad hoc network according to claim 1, characterized in that, Step S1 includes: Perform timestamp alignment and noise reduction cleaning operations based on linear interpolation on the acquired multidimensional sensor data stream to obtain synchronized and cleaned sensor data; Spatial mapping calculations involving coordinate translation and rotation transformations are performed on the sensor data after synchronous cleaning to obtain the bucket wheel end pose data; The velocity components in the bucket wheel end pose data are estimated using a Kalman filter. The calculated global position coordinates, filtered linear velocity, estimated acceleration, and attitude angle data are then assembled into vectors according to a preset communication protocol to obtain the current state vector of the machine.

3. The multi-node cooperative control method for bucket wheel excavators based on a wireless ad hoc network according to claim 1, characterized in that, Step S2 includes: The current state vector of the machine and the target position parsed in the job task configuration are used to perform path planning and time-domain discretization sampling to obtain a discrete reference path point set that meets the preset geometric boundary conditions. Using a discrete reference path point set as the tracking target, a quadratic programming solver is used to perform finite-time domain optimization calculations to obtain the nominal control input sequence; Starting from the current state vector of the local machine, a forward state recursion calculation is performed on the nominal control input sequence to generate the desired trajectory sequence of the local machine.

4. The multi-node cooperative control method for bucket wheel excavators based on a wireless ad hoc network according to claim 1, characterized in that, Step S3 includes: The received neighbor node broadcast data stream is deframed according to the protocol to obtain the set of neighbor geometric trajectories; The information age index is calculated by generating timestamps from data parsed from the broadcast data stream of neighboring nodes, and then combined with the link quality indication in the normalized physical layer signal characteristics for a comprehensive weighted calculation of timeliness and signal quality to generate a link reliability score set. Using the link reliability score set, a spatial sphere expansion process based on dynamic safety radius is performed on each predicted location point in the neighbor geometric trajectory set to obtain a weighted neighbor trajectory set.

5. The multi-node cooperative control method for bucket wheel excavators based on a wireless ad hoc network according to claim 1, characterized in that, Step S4 includes: Based on the dynamic safety radius and predicted location information contained in the weighted neighbor trajectory set, the spatial geometric relationship between the local machine and neighboring nodes in the prediction time domain is modeled using the Euclidean distance inequality to obtain the collision detection constraint set. With the goal of minimizing the tracking deviation from the desired trajectory sequence of the local machine, the collision detection constraint set is used as the boundary condition of the nonlinear programming problem and input into the nonlinear model predictive control solver for constraint optimization iterative calculation to obtain the optimal control input sequence. Using the current state vector of the machine as the initial condition, the optimal control input sequence is simulated and replayed in the forward direction, and the state data is encapsulated to obtain the optimal safe trajectory sequence.

6. The multi-node cooperative control method for bucket wheel excavators based on a wireless ad hoc network according to claim 1, characterized in that, Step S5 includes: The target state point for the next moment is extracted from the optimal safe trajectory sequence, and the deviation vectorization calculation based on the feedforward gain matrix is ​​performed in combination with the current state vector of the machine to generate the instantaneous control deviation vector. The instantaneous control deviation vector is used to perform control law calculations to generate the original motor drive signal; Based on the fieldbus protocol, the original motor drive signal is mapped to the controller output register address and sent to the trolley frequency converter and slewing hydraulic station through the physical hardware interface to drive the actuator to move, thereby outputting the execution action command.

7. A multi-node collaborative control system for a bucket wheel excavator based on a wireless ad hoc network, characterized in that, include: The time alignment and coordinate mapping solution module is used to perform time alignment and coordinate mapping solution on the acquired multi-dimensional sensor data stream of the bucket wheel excavator to obtain the current state vector of the excavator containing real-time position, velocity and acceleration information; The forward planning module is used to perform forward planning on the current state vector of the machine within a set prediction time domain according to the preset task configuration, so as to generate the expected trajectory sequence of the machine containing multiple discrete time steps in the future. The weighted neighbor trajectory set acquisition module is used to parse the compressed trajectory data broadcast by the received neighbor nodes to obtain the neighbor node trajectory sequence, calculate the link reliability coefficient based on the physical layer signal characteristics, and use the link reliability coefficient to perform channel quality weighting on the neighbor node trajectory sequence to obtain the weighted neighbor trajectory set. The spatiotemporal conflict detection and constraint optimization solution module is used to perform spatiotemporal conflict detection and constraint optimization on the local desired trajectory sequence and the weighted neighbor trajectory set to obtain the optimal safe trajectory sequence. The closed-loop deviation feedback calculation module is used to perform closed-loop deviation feedback calculation on the optimal safe trajectory sequence based on the current state vector of the machine, so as to obtain the execution action commands for driving the bucket wheel excavator trolley travel and cantilever slewing mechanism.

8. The multi-node collaborative control system for bucket wheel excavators based on a wireless ad hoc network according to claim 7, characterized in that, The time alignment and coordinate mapping solution module includes: The timestamp alignment and denoising cleaning unit is used to perform timestamp alignment and denoising cleaning operations based on linear interpolation on the acquired multidimensional sensor data stream to obtain synchronized and cleaned sensor data. The spatial mapping solution unit is used to perform spatial mapping solution, including coordinate translation and rotation transformation, on the sensor data after synchronous cleaning to obtain the bucket wheel end pose data. The current state vector acquisition unit is used to estimate the acceleration of the velocity component in the bucket wheel end pose data using a Kalman filter, and to assemble the calculated global position coordinates, filtered linear velocity, estimated acceleration and attitude angle data into vectors according to a preset communication protocol order to obtain the current state vector of the machine.

9. The multi-node collaborative control system for bucket wheel excavators based on a wireless ad hoc network according to claim 7, characterized in that, The module for obtaining the weighted nearest neighbor trajectory set includes: The protocol deframing unit is used to perform protocol deframing on the received neighbor node broadcast data stream to obtain a set of neighbor geometric trajectories; The link reliability score set generation unit is used to generate a timestamp based on the data parsed from the broadcast data stream of neighboring nodes, calculate the information age index, and perform a comprehensive weighted calculation of timeliness and signal quality with the link quality indication in the normalized physical layer signal characteristics to generate a link reliability score set. The spatial sphere expansion processing unit is used to perform spatial sphere expansion processing based on dynamic safety radius on each predicted location point in the neighbor geometric trajectory set using the link reliability score set, so as to obtain a weighted neighbor trajectory set.

10. The multi-node collaborative control system for bucket wheel excavators based on a wireless ad hoc network according to claim 7, characterized in that, The spatiotemporal conflict detection and constraint optimization solution module includes: The Euclidean distance inequality modeling unit is used to model the spatial geometric relationship between the local machine and its neighboring nodes in the prediction time domain based on the dynamic safety radius and predicted location information contained in the weighted neighbor trajectory set, so as to obtain the collision detection constraint set. The optimal control input sequence acquisition unit is used to minimize the tracking deviation from the local desired trajectory sequence as the optimization objective. It inputs the collision detection constraint set as the boundary condition of the nonlinear programming problem into the nonlinear model predictive control solver for constraint optimization iterative calculation to obtain the optimal control input sequence. The optimal safety trajectory sequence acquisition unit is used to perform forward simulation playback and state data encapsulation on the optimal control input sequence with the current state vector of the local machine as the initial condition to obtain the optimal safety trajectory sequence.

Images