Intelligent cooperative control method for wind and fog of boom sprayer based on digital twinning

Abstract

The invention discloses a boom sprayer wind and fog intelligent cooperative control method based on digital twinning, and the method comprises the steps: collecting multi-source operation state data, and generating standardized multi-source operation state data; constructing an asynchronous state fusion module, and generating a continuous time digital twinning state; carrying out component credibility modulation to generate a credibility-enhanced wind-fog coupling digital twinning state; extracting a wind-fog coupling key state index to form an expected wind-fog cooperative control quantity; dynamically adjusting a cooperative relation structure of the fan and spraying control parameters, and generating a structure-adaptive air-mist cooperative control quantity; and constructing an improved BoucWen model, and outputting fan control input and spray control input. According to the invention, by constructing the asynchronous state fusion module and improving the BoucWen model, stable perception and accurate cooperative adjustment of the boom sprayer on the wind and fog state under the condition of a complex wind field are realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for agricultural equipment, and in particular to an intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins. Background Technology

[0002] Currently, with the continuous improvement of agricultural mechanization and intelligence, boom sprayers are increasingly widely used in the prevention and control of crop diseases and pests. The quality of spraying operations has a significant impact on pesticide utilization efficiency and operational safety. During spraying operations, the deposition effect and drift of spray droplets are affected by a variety of factors, including changes in natural wind fields, spray parameter settings, and operating conditions, making the operating environment dynamic and uncertain. In traditional technologies, most boom sprayers still rely on fixed parameters or manual experience for spray control, making it difficult to adjust the spray state in real time according to changes in the wind field during operation. This easily leads to problems such as spray drift, uneven coverage, or pesticide waste, making it difficult to meet the actual needs of precision plant protection operations.

[0003] While existing technologies attempt to incorporate sensor data or control models to regulate the spraying process, they are mostly based on discrete-time modeling or single-state variables. This makes it difficult to fully integrate multi-source heterogeneous data such as wind field conditions, fan airflow conditions, and spray conditions, resulting in insufficient characterization of the continuously changing wind-mist coupling relationship during operation. In the coordinated regulation of fan control and spray control, existing methods generally lack a systematic description of their dynamic coupling relationship, making it difficult to achieve stable and coordinated control effects under complex wind field conditions. They also have limitations in terms of model continuity, coordination, and the expression of control response hysteresis characteristics.

[0004] Therefore, how to provide a method for intelligent collaborative control of wind and mist in boom sprayers based on digital twins is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a digital twin-based intelligent collaborative control method for boom sprayers, specifically for wind and mist. This invention fully utilizes multi-source operational state perception technology, continuous-time state modeling methods, and collaborative control modeling concepts to construct a continuous-time digital twin model of the wind field state, fan airflow state, and spray state during boom sprayer operation. It details the dynamic expression and collaborative adjustment process of the wind-mist coupling relationship under complex wind field conditions. Through credibility modulation of the wind-mist coupling state, extraction of key state indicators, and short-time state extrapolation, adaptive collaborative optimization of fan control parameters and spray control parameters is achieved. Combined with an improved BoucWen model, executable control commands are output. This invention effectively improves the adaptability of spraying operations to wind field changes, reduces spray drift risk, and enhances spray coverage uniformity and operational stability. It boasts advantages such as high control accuracy, strong collaboration, and good operational robustness.

[0006] A method for intelligent collaborative control of wind and mist in a boom sprayer based on digital twins according to an embodiment of the present invention includes: Collect multi-source operation status data of boom sprayers, preprocess the multi-source operation status data, and generate standardized multi-source operation status data; An asynchronous state fusion module based on neural controlled differential equations is constructed to perform continuous-time modeling and state evolution expression of standardized multi-source operation state data with different time scales and sampling frequencies, generating continuous-time digital twin states between wind field state, fan airflow state and spray state. Based on the continuous-time digital twin state, component credibility modulation is performed to generate a wind-fog coupled digital twin state with enhanced credibility. Based on the enhanced credibility of the wind-fog coupled digital twin state, key state indicators of wind-fog coupling are extracted, and short-term state simulation is performed to obtain the evolution trend of wind-fog state during spraying operations, and to form the desired wind-fog coordinated control quantity. Based on the expected wind-fog coordinated control quantity, according to the evolution trend of key state indicators of wind-fog coupling and the actual control response deviation, the coordinated relationship structure between the fan control parameters and the spray control parameters is dynamically adjusted to generate a structurally adaptive wind-fog coordinated control quantity. An improved BoucWen model is constructed to perform response mapping on the structurally adaptive wind and fog coordinated control variables, generate corresponding fan hysteresis state variables and spray hysteresis state variables, and output the fan control input and spray control input that can be actually executed.

[0007] Optionally, the multi-source operation status data specifically includes wind field status data, fan operation status data, sprayer operation status data, boom operation status data, and operation parameter data.

[0008] Optionally, the preprocessing of multi-source operation status data specifically includes time consistency processing, abnormal and missing data processing, unit and scale unification processing, noise suppression processing, and data validity labeling.

[0009] Optionally, the generation of a continuous-time digital twin state between the wind field state, the fan airflow state, and the spray state includes: An asynchronous state fusion module based on neural controlled differential equations is constructed. The asynchronous state fusion module consists of a data time processing unit, a continuous time driving unit, a state evolution unit, an observation fusion update unit, and a fusion state output unit. Standardized multi-source operation status data is input into the data time processing unit, and timestamp parsing, time sorting and time alignment are performed on each data component to form an observation sequence arranged in chronological order, and to generate a time interval sequence and a missing measurement marker sequence between adjacent observation times. The observation sequence, time interval sequence, and missing data identification sequence are input into the continuous-time driving unit to construct an adjustable multi-frequency control curve superposition structure. This generates a low-frequency control curve that represents the long-term trend and a high-frequency control curve that represents the short-term disturbance characteristics. These are then weighted and superimposed to form a continuous-time control curve. The continuous-time control curve is input into the state evolution unit, the continuous-time hidden state is set as the internal fusion state, a state update adjustment gate is introduced, and the update amplitude of the continuous-time hidden state is adjusted according to the change characteristics of the continuous-time control curve. A dual-pathway collaborative evolution structure is constructed, and the main path state channel and the offset path state channel are continuously updated in parallel along the time axis under the drive of the neural controlled differential equation to obtain the continuous-time hidden state. At each observation time, the observation vector, missing data identifier sequence and time interval sequence are input into the observation fusion update unit. The observation fusion update unit performs fusion update on the effective observation components under the constraint of the missing data identifier sequence, and adjusts the fusion update amplitude according to the time interval sequence to form the fusion hidden state at the observation time. The fused hidden state is input into the fused state output unit and introduced into the continuous state spectrum decomposition channel. The fused hidden state is subjected to spectrum decomposition processing along the time dimension. The state components corresponding to the wind field state, the fan airflow state and the spray state are frequency-projected and combined respectively to obtain the frequency-aware continuous time state components and generate a continuous time digital twin state.

[0010] Optionally, the generation of the enhanced credibility wind-fog coupled digital twin state includes: The continuous-time digital twin state is divided into wind field state components, fan airflow state components, and spray state components, forming a component set; Based on the component set, state consistency determination is performed for each state component. The direction and magnitude of continuous change of each state component between adjacent observation times are matched to generate the corresponding consistency deviation. Based on the consistency deviation, physical accessibility is determined for each state component. The accessibility range of the wind field state component, the fan airflow state component, and the spray state component is limited according to the operating parameter data of the boom sprayer, and the accessibility deviation is obtained. Based on the consistency deviation and reachability deviation, component confidence values ​​are generated for each state component. Then, based on the component confidence values, weighted fusion and consistency verification are performed on the wind field state component, the fan airflow state component, and the spray state component to obtain a wind-fog coupled digital twin state with enhanced confidence.

[0011] Optionally, the formation of the desired wind-fog coordinated control quantity includes: Based on the enhanced credibility of the wind-fog coupled digital twin state, a wind-fog coupled state sequence consisting of wind field state components, fan airflow state components and spray state components is constructed in chronological order. Based on the wind-fog coupled state sequence, the wind field state component, the fan airflow state component and the spray state component are jointly analyzed to extract key wind-fog coupled state indicators that characterize the wind-fog collaborative operation, forming a set of key wind-fog coupled state indicators. Based on the set of key state indicators of wind and fog coupling, a short-term state simulation window is set, and the time evolution relationship of key state indicators of wind and fog coupling is simulated within the short-term state simulation window to obtain the evolution sequence of key state indicators of wind and fog coupling within the short-term state simulation window. Based on the evolution sequence, a mapping relationship between key state indicators of wind-fog coupling and control parameters is constructed, and the target value range of wind turbine control parameters and spray control parameters is determined. Based on the target value ranges of the fan control parameters and the spray control parameters, the control parameters are matched and conflict resolution is performed to generate the desired wind and fog coordinated control quantity.

[0012] Optionally, the wind and fog cooperative control quantity generated by the structure includes: The desired wind-fog coordinated control quantity is decomposed into the initial vector of the fan control parameters and the initial vector of the spray control parameters, and a coordinated relationship map is constructed based on the evolution trend of the key state indicators of wind-fog coupling. In the collaborative relationship graph, based on the evolution trend of key state indicators of wind and fog coupling, the initial weights of the coupling strength of each side are updated with trend consistency to obtain the trend update weights, and a collaborative relationship graph reflecting the evolution trend is formed. The actual control response results after executing the fan control input and spray control input are obtained. Based on the deviation between the actual control response results and the evolution trend of the key state indicators of wind and fog coupling, a response deviation amount is generated. Based on the response deviation amount, the trend update weights of each side are updated to perform deviation correction and obtain the deviation correction weight. Based on the deviation correction weights, the cooperative relationship graph is structurally adaptively adjusted to generate a structurally adaptive cooperative relationship graph. Based on the structurally adaptive cooperative relationship graph, the initial vectors of the fan control parameters and the initial vectors of the spray control parameters are subjected to cooperative reorganization and consistency verification under the constraints of the cooperative relationship graph, and the structurally adaptive wind and fog cooperative control quantity is output.

[0013] Optionally, the output may include actual executable fan control inputs and spray control inputs, including: An improved Bouc-Wen model is constructed, which consists of a dual-pathway hysteresis unit, a gating regulation unit, a memory enhancement unit, and a response fusion unit. The adaptive wind and fog coordinated control quantity is decomposed into a fan control parameter vector and a spray control parameter vector, and then input into the fan path and spray path of the dual-path hysteresis unit respectively to form a fan hysteresis drive sequence and a spray hysteresis drive sequence. The gating control unit generates fan gating quantity and spray gating quantity respectively based on the change amplitude and change direction of the fan hysteresis drive sequence and the spray hysteresis drive sequence, introduces an asymmetric gating response path, and differentially adjusts the hysteresis evolution weight of the fan path and the spray path. In the dual-path hysteresis unit, a segmented hysteresis evolution path is introduced. Based on the hysteresis evolution weights of the fan gating quantity, the spray gating quantity, and the fan path and spray path, hysteresis evolution processing is performed on the fan hysteresis drive sequence and the spray hysteresis drive sequence. The drive sequence is divided into two response intervals, and the fan hysteresis state quantity and the spray hysteresis state quantity are updated in each response interval to obtain the segmented updated fan hysteresis state quantity and spray hysteresis state quantity. The updated fan hysteresis state and spray hysteresis state are input into the memory enhancement unit and introduced into the inertial control sub-channel. The current hysteresis state is combined with the historical hysteresis state to perform inertial adjustment, so as to obtain the fan hysteresis state and spray hysteresis state after inertial adjustment. The inertial-adjusted fan hysteresis state quantity and the inertial-adjusted spray hysteresis state quantity are input into the response fusion unit. The state disturbance suppression filtering process is performed on the structure-adaptive wind and fog coordinated control quantity to suppress high-frequency disturbance components. The filtered control quantity is then fused with the corresponding hysteresis state quantity to generate the fan control input and the spray control input.

[0014] The beneficial effects of this invention are: This invention proposes a digital twin-based intelligent collaborative control method for boom sprayers, integrating multi-source operational state perception, continuous-time state modeling, and collaborative control modeling techniques to systematically model and represent the dynamic coupling relationship between wind field state, fan airflow state, and spray state during boom sprayer operation. By performing continuous-time modeling on multi-source operational state data, a digital twin state reflecting the wind and fog evolution process is constructed, achieving a unified expression of asynchronous, multi-scale operational states. A component reliability modulation mechanism is introduced to assess and enhance the reliability of the wind and fog coupling state, and key state indicators are further extracted to extrapolate the evolution trend of the wind and fog state within a short operational period, forming a forward-looking basis for collaborative control. By constructing a collaborative relationship model between fan control parameters and spray control parameters, dynamic coordination between the two under complex wind field conditions is achieved. Combined with an improved BoucWen model, the collaborative control results are mapped into executable control inputs, completing the closed-loop regulation of wind and fog collaborative control. This invention can accurately characterize the continuous change characteristics of wind and fog state during spraying operations, overcoming the problem that traditional discrete modeling and static parameter adjustment are difficult to adapt to complex operating environments, and realizing collaborative optimization between fan control and spray control. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0016] Figure 1 The flowchart shows a method for intelligent collaborative control of wind and mist in a boom sprayer based on digital twin proposed in this invention. Figure 2 This is a schematic diagram of the asynchronous state fusion module structure of the intelligent collaborative control method for wind and mist of a boom sprayer based on digital twin proposed in this invention; Figure 3 This is a schematic diagram of the improved BoucWen model of the intelligent collaborative control method for wind and mist of a boom sprayer based on digital twin proposed in this invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0018] refer to Figure 1 , Figure 2 and Figure 3 A method for intelligent collaborative control of wind and mist in a boom sprayer based on digital twins, comprising: Collect multi-source operation status data of boom sprayers, preprocess the multi-source operation status data, and generate standardized multi-source operation status data; An asynchronous state fusion module based on neural controlled differential equations is constructed to perform continuous-time modeling and state evolution expression of standardized multi-source operation state data with different time scales and sampling frequencies, generating continuous-time digital twin states between wind field state, fan airflow state and spray state. Based on the continuous-time digital twin state, component credibility modulation is performed to generate a wind-fog coupled digital twin state with enhanced credibility. Based on the enhanced credibility of the wind-fog coupled digital twin state, key state indicators of wind-fog coupling are extracted, and short-term state simulation is performed to obtain the evolution trend of wind-fog state during spraying operations, and to form the desired wind-fog coordinated control quantity. Based on the expected wind-fog coordinated control quantity, according to the evolution trend of key state indicators of wind-fog coupling and the actual control response deviation, the coordinated relationship structure between the fan control parameters and the spray control parameters is dynamically adjusted to generate a structurally adaptive wind-fog coordinated control quantity. An improved BoucWen model is constructed to perform response mapping on the structurally adaptive wind and fog coordinated control variables, generate corresponding fan hysteresis state variables and spray hysteresis state variables, and output the fan control input and spray control input that can be actually executed.

[0019] In this embodiment, the multi-source operation status data specifically includes wind field status data, wind turbine operation status data, sprayer operation status data, boom operation status data, and operation parameter data.

[0020] In this embodiment, the preprocessing of multi-source operation status data specifically includes time consistency processing, abnormal and missing data processing, unit and scale unification processing, noise suppression processing, and data validity labeling.

[0021] In this embodiment, the generation of a continuous-time digital twin state between the wind field state, the fan airflow state, and the spray state includes: An asynchronous state fusion module based on neural controlled differential equations is constructed. This module comprises a data time processing unit, a continuous-time driving unit, a state evolution unit, an observation fusion update unit, and a fusion state output unit. Specifically, the construction of this asynchronous state fusion module based on neural controlled differential equations is as follows: The data time processing unit, continuous time driving unit, state evolution unit, observation fusion update unit, and fusion state output unit are sequentially connected and work together to form an asynchronous state fusion module based on neural controlled differential equations; Standardized multi-source operational status data is input into the data time processing unit. Timestamp parsing, time sorting, and time alignment are performed on each data component to form an observation sequence arranged chronologically. A time interval sequence and a missing data marker sequence are also generated between adjacent observation times. Specifically, the timestamp parsing, time sorting, and time alignment processing for each data component involves: The time stamp carried by each data component is read and parsed, and time stamps from different sources and in different formats are uniformly converted into a standard time expression. Based on the parsed time stamps, the data components are sorted in chronological order to form a data sequence arranged in ascending order of time. Based on a unified time reference, time alignment processing is performed on the sorted data components to map each data component to the corresponding observation time. Data components that were not collected at the corresponding observation time are marked as missing data. During the time alignment process, the time difference between adjacent observation times is calculated to generate a time interval sequence, and a missing data indicator sequence reflecting whether each data component is missing is generated simultaneously. The observation sequence, time interval sequence, and missing data marker sequence are input into the continuous-time drive unit to construct an adjustable multi-frequency control curve superposition structure. This generates low-frequency control curves characterizing long-term trends and high-frequency control curves characterizing short-term disturbances, which are then weighted and superimposed to form the continuous-time control curve. The construction of the adjustable multi-frequency control curve superposition structure is specifically as follows: Based on the state change amplitude corresponding to the observation sequence, the sampling rhythm reflected by the time interval sequence, and the data integrity reflected by the missing measurement indicator sequence, the observation sequence is divided into multiple time scales, and control curve channels corresponding to different change rates are constructed. The initial weight parameters of each control curve channel are set according to the statistical characteristics of the state change amplitude within the corresponding time scale. The sampling interval change is evaluated according to the time interval sequence, and the data integrity is judged in combination with the missing measurement indicator sequence. Adaptive correction processing is performed on the weight parameters of each control curve channel. When the sampling interval increases and the missing measurement ratio increases, the weight of the corresponding high-frequency control curve channel is reduced. When the sampling interval decreases and the data integrity is high, the weight of the corresponding high-frequency control curve channel is increased. The generation of low-frequency control curves characterizing long-term trends and high-frequency control curves characterizing short-term disturbances, respectively, is as follows: Based on the state change characteristics that persist within five consecutive observation times in the observation sequence, the observation sequence is stretched on a time scale, and adjacent observation data are smoothly aggregated by combining the time interval sequence to form a continuous and smooth low-frequency control curve. Based on the change characteristics between adjacent observation times in the observation sequence, the state change rate is characterized by combining the time interval sequence, and rapid change extraction is performed only on the effective observation components under the constraint of the missing measurement label sequence, forming a high-frequency control curve that characterizes the short-term disturbance characteristics. The continuous-time control curve is input into the state evolution unit, and the continuous-time hidden state is set as the internal fusion state. A state update adjustment gate is introduced, and the update amplitude of the continuous-time hidden state is adjusted according to the change characteristics of the continuous-time control curve. A dual-pathway co-evolution structure is constructed. Under the drive of the neural controlled differential equation, the main path state channel and the offset path state channel are continuously updated in parallel along the time axis to obtain the continuous-time hidden state, where: The continuous-time hidden state refers to taking the observation sequence corresponding to the earliest observation time as the initial state source, performing state mapping processing on the wind field state components, wind turbine airflow state components and spray state components contained in the observation sequence, and mapping the discrete observation components into a set of state variables in a unified state space to form a continuous-time hidden state. The state update regulation gate is a state modulation component embedded in the state evolution unit. It describes the structural characteristics of the change of update intensity of the continuous-time hidden state with time during the evolution process. It serves as a structural connection interface between the continuous-time control curve and the continuous-time hidden state evolution, and describes the difference in the update amplitude of the hidden state participating in the state evolution in different time periods. The construction of the dual-pathway collaborative evolution structure specifically involves: The evolution channels of the continuous-time hidden state are divided into the main path state channel and the offset path state channel. The main path state channel carries the stable evolution state component dominated by the continuous-time control curve, while the offset path state channel carries the offset state component generated relative to the main path evolution. The two state channels are structurally independent but share the same continuous-time hidden state space and coexist under the same time parameters, forming a dual-path parallel state evolution structure. The parallel and continuous updating of the main path state channel and the offset path state channel along the time axis under the drive of neural controlled differential equations is specifically as follows: The continuous-time control curve is regarded as a driving input signal continuously defined between adjacent observation times, and the continuous-time control curve is used as the time continuous input of state evolution. In each small time increment, the hidden state components corresponding to the main path state channel and the offset path state channel are calculated once. The main path state channel updates the hidden state components of the main path step by step based on the current value of the continuous-time control curve, forming a continuous state trajectory of the main path covering adjacent observation times. The offset path state channel updates the hidden state components of the offset path synchronously at the same time step, forming a continuous state trajectory of the offset path that corresponds one-to-one with the main path state trajectory in time. At each consecutive time position, the hidden state components obtained by the main path state channel and the offset path state channel at the time position are jointly recorded to form the joint hidden state vector at the corresponding time. By continuously arranging the joint hidden state vectors at each time position in the continuous time interval, the continuous time hidden state describing the state change process in the corresponding time interval is obtained. At each observation time, the observation vector, the missing data identifier sequence, and the time interval sequence are input into the observation fusion update unit. The observation fusion update unit performs fusion updates on the valid observation components under the constraint of the missing data identifier sequence, and adjusts the fusion update amplitude according to the time interval sequence to form the fused hidden state at the observation time. Specifically, the fusion update of the valid observation components under the constraint of the missing data identifier sequence is performed by the observation fusion update unit as follows: At each observation time, the corresponding observation vector is aligned and matched with the current continuous-time hidden state. The validity of each observation component in the observation vector is determined based on the missing data identifier sequence. Observation components marked as missing data do not participate in the fusion update process. For observation components marked as valid, the corresponding observation values ​​are extracted. The valid observation components are then fused with the continuous-time hidden state components obtained from the state evolution unit at the observation time. The time interval between the current observation time and the previous observation time is obtained based on the time interval sequence and used as the basis for adjusting the update intensity. The update amplitude during the fusion update process is adjusted. When the time interval between adjacent observation times is greater than the interval threshold, the correction amplitude of the observation to the hidden state is reduced; when the time interval between adjacent observation times is less than the interval threshold, the correction amplitude of the observation to the hidden state is increased, forming the fused hidden state. The fusion calculation is specifically performed as follows: The hidden state component is used as the prior state input, and the valid observation component is used as the correction information input. For each pair of observation components and hidden state components in the matching set, state correction processing is performed. The hidden state component is adjusted in the direction of the observation value while maintaining the original continuous evolution trend. The observation component marked as invalid by the missing detection sequence does not participate in the state correction processing, and the corresponding hidden state component remains unchanged. For the observation component marked as valid, the corresponding state correction amount is calculated according to the magnitude of the deviation between the corresponding hidden state component and the observation component, and the state correction amount is superimposed on the hidden state component. The fused hidden state is input into the fused state output unit and introduced into the continuous state spectrum decomposition channel. Spectral decomposition processing is performed on the fused hidden state along the time dimension. Frequency projection and combination are performed on the state components corresponding to the wind field state, fan airflow state, and spray state, respectively, to obtain frequency-aware continuous-time state components, generating a continuous-time digital twin state, where: The continuous state spectrum decomposition channel is a state expression reconstruction channel embedded in the fusion state output unit. It describes the composition structure of the continuous time hidden state in the time dimension. After fusion, the hidden state is organized into a state expression containing components with different frequency changes. The continuous state spectrum decomposition channel decomposes and reorganizes the structure of the hidden state changing with time in a unified state space, characterizes the changing characteristics of wind field state, wind turbine airflow state and spray state at different time scales, and forms a frequency-sensing expression of the dynamic characteristics of the state. The generation of continuous-time digital twin states specifically involves: The fused hidden states obtained at the observation time are used as the input state set. A continuous state spectrum decomposition channel is introduced to reconstruct the spectrum structure of the fused hidden states in the continuous time dimension. Each hidden state component is mapped to the corresponding frequency expression space. Based on the correspondence between the hidden state components and the wind field state, the fan airflow state, and the spray state, frequency projection and combination processing are performed on the decomposed frequency expression results. Different frequency components are remapped into state components with physical semantics. By continuously arranging the frequency-sensing state components at each time position on the continuous time axis, a continuous-time digital twin state that simultaneously contains temporal continuity and frequency structure information is finally formed.

[0022] In this embodiment, generating a more reliable wind-fog coupled digital twin state includes: The continuous-time digital twin state is divided into wind field state components, fan airflow state components, and spray state components, forming a component set; Based on the component set, a state consistency determination is performed for each state component. The direction and magnitude of continuous change of each state component between adjacent observation times are matched to generate a corresponding consistency deviation. Specifically, the state consistency determination for each state component involves: For each state component, the value changes of the state component at adjacent observation times are extracted, the direction of change between adjacent observation times is compared and determined, and the corresponding change amplitude is quantified and calculated. The actual change direction of the current state component between adjacent observation times is matched with the change direction in the continuous time hidden state evolution process. At the same time, the actual change amplitude is aligned and compared with the corresponding continuous change amplitude. When the change direction is consistent and the change amplitude is within the matching range threshold, the state component is determined to have high consistency in the observation interval. When the change direction is inconsistent and the change amplitude deviates from the matching range threshold, the corresponding direction deviation and amplitude deviation are recorded, and the direction deviation and amplitude deviation are combined to form the consistency deviation of the state component between adjacent observation times. Based on the consistency deviation, physical accessibility is determined for each state component. The accessibility ranges of the wind field state component, the fan airflow state component, and the spray state component are limited according to the operating parameters of the boom sprayer, resulting in the accessibility deviation. Specifically, the physical accessibility determination for each state component involves: For each state component, the corresponding operating information in the operating parameter data of the boom sprayer is read, and a corresponding physical reachability range is set for the state component based on the operating information. The value change of the state component between the current observation time and adjacent observation times is compared with the corresponding physical reachability range. When the value and change range of the state component are both within the physical reachability range, it is determined that the state component meets the physical reachability requirements within the observation interval. When the value of the state component exceeds the physical reachability range, the corresponding degree of exceeding the limit is recorded. After completing the determination of each state component, the recorded degree of exceeding the limit is organized to form the corresponding reachability deviation. Based on the consistency deviation and reachability deviation, component confidence values ​​are generated for each state component. Then, based on these confidence values, weighted fusion and consistency verification are performed on the wind field state component, the fan airflow state component, and the spray state component to obtain a more reliable wind-fog coupled digital twin state. Specifically, obtaining the more reliable wind-fog coupled digital twin state involves: For each state component, the consistency deviation and reachability deviation are normalized and weighted to generate a corresponding component confidence value. The component confidence value is then correlated with the corresponding wind field state component, fan airflow state component, and spray state component. The component confidence value is used as the weighting basis to perform weighted fusion processing on each state component. After weighted fusion, the coupling relationship between the wind field state component, fan airflow state component, and spray state component in the fusion result is verified for consistency. Specifically, the direction, magnitude, and mutual constraints of change of each fused state component within the current observation interval are matched and judged. When the change relationship between the state components is consistent, the fusion result is determined to be consistent within the observation interval, and the fused state result is output, forming a wind-fog coupled digital twin state with enhanced confidence.

[0023] In this embodiment, the formation of the desired wind and fog coordinated control quantity includes: Based on the enhanced credibility of the wind-fog coupled digital twin state, a wind-fog coupled state sequence consisting of wind field state components, fan airflow state components and spray state components is constructed in chronological order. Based on the wind-fog coupled state sequence, the wind field state component, the fan airflow state component, and the spray state component are jointly analyzed to extract key wind-fog coupled state indicators characterizing the characteristics of wind-fog coordinated operation, forming a set of key wind-fog coupled state indicators. Specifically, the extraction of these key wind-fog coupled state indicators characterizing the characteristics of wind-fog coordinated operation includes: At each observation time, the corresponding changes in the wind field state components and the spray state components before and after the wind field state components change are calculated to form the correspondence between wind field changes and wind turbine airflow response and spray state changes. Between adjacent observation times, the change amplitude and change rate of each state component are statistically analyzed. The corresponding change relationship, change amplitude, and change rate are combined to generate key state indicators of wind-fog coupling. Based on the set of key state indicators coupled with wind and fog, a short-term state simulation window is set, and the temporal evolution relationship of the key state indicators coupled with wind and fog is simulated within the short-term state simulation window to obtain the evolution sequence of the key state indicators coupled with wind and fog within the short-term state simulation window, wherein: The setting of the short-term state simulation window is specifically as follows: Based on the time series corresponding to the key state indicators of wind and fog coupling, three consecutive time intervals are extracted before and after the current observation time as short-term state inference windows. The process of extrapolating the temporal evolution of key state indicators coupled with wind and fog within a short-term state extrapolation window is as follows: The short-term state simulation window is divided into a continuous simulation time sequence. At each simulation time, the key state index value corresponding to the previous simulation time is read, and the change increment of the key state index at the current simulation time is determined based on the change results of adjacent observation times in the wind-fog coupled state sequence. The change increment is superimposed on the key state index value of the previous simulation time to obtain the key state index value at the current simulation time. When the missing measurement indicator sequence indicates that a certain state component is missing, the change increment is calculated only from the change results corresponding to the effective state component. This process is repeated until the simulation time sequence ends, forming an evolution sequence. Based on the evolutionary sequence, a mapping relationship between key state indicators and control parameters of wind-mist coupling is constructed, and the target value ranges of fan control parameters and spray control parameters are determined. Specifically, determining the target value ranges of fan control parameters and spray control parameters involves: Based on the evolution sequence of key state indicators of wind and fog coupling within a short-term state simulation window, the range and trend of value changes of each key state indicator within the window are analyzed. Based on the correspondence between key state indicators and fan control parameters and spray control parameters, the evolution results of key state indicators are mapped to the adjustable range of control parameters, and the target value range of fan control parameters and spray control parameters is obtained. Based on the target value ranges of the fan control parameters and the spray control parameters, collaborative matching and conflict resolution are performed on each control parameter to generate the desired wind-mist collaborative control quantity. Specifically, the collaborative matching and conflict resolution process for each control parameter involves: Based on the synergistic relationship reflected by the key state indicators of wind-fog coupling, parameter combination matching is performed on the fan control parameters and spray control parameters. Parameter combinations that satisfy the synergistic relationship constraints are selected. When there are conflicting control actions or situations exceeding the synergistic relationship constraints in the parameter combinations, conflicting parameters are adjusted or discarded according to the priority relationship of each control parameter in wind-fog synergistic operation to eliminate conflicts between parameters. After completing the synergistic matching and conflict resolution, the remaining parameter combinations are aggregated to generate the desired wind-fog synergistic control quantity, where: The coordination constraint means that the adjustment direction and adjustment range of the fan control parameters must be coordinated with the adjustment direction and adjustment range of the spray control parameters, and the fan control parameters and spray control parameters must meet the physical boundaries of the operation parameter data.

[0024] In this embodiment, the wind and fog cooperative control quantity that generates the adaptive structure includes: The desired wind-fog coordinated control quantity is decomposed into the initial vector of wind turbine control parameters and the initial vector of spray control parameters. Based on the evolution trend of key state indicators of wind-fog coupling, a coordinated relationship graph is constructed. The construction of the coordinated relationship graph specifically involves: Using the fan control parameters and spray control parameters as nodes in the graph, and based on the evolution trend of key state indicators of wind-mist coupling, the influence relationship of changes in fan control parameters on changes in spray state and the feedback relationship of changes in spray control parameters on fan airflow response are analyzed. The association edges between the fan control parameter nodes and the spray control parameter nodes are established, and the linkage direction and linkage intensity reflected by the evolution trend are used as edge attributes to form a collaborative relationship graph. In the collaborative relationship graph, based on the evolution trend of key state indicators of wind-fog coupling, the initial weights of the coupling strength of each edge are updated with trend consistency to obtain trend-updated weights, thus forming a collaborative relationship graph that reflects the evolution trend. Specifically, the update of the initial weights of the coupling strength of each edge with trend consistency is performed as follows: In the collaborative relationship graph, the evolution trend of the key state indicators of wind and fog coupling corresponding to the control parameters associated with the edge is read, and the evolution trend is compared with the parameter linkage direction. When the evolution trend of the key state indicator is consistent with the parameter linkage direction, the initial weight of the coupling strength corresponding to the edge is increased. When the evolution trend of the key state indicator is inconsistent with the parameter linkage direction, the initial weight of the coupling strength corresponding to the edge is decreased to obtain the updated weight and form the corresponding collaborative relationship graph. The actual control response results after executing the fan control input and spray control input are obtained. Based on the deviation between the actual control response results and the evolution trend of the key state indicators of wind-mist coupling, a response deviation is generated. Based on the response deviation, deviation correction is performed on the trend update weights of each side to obtain the deviation correction weights. Specifically, the deviation correction based on the response deviation is performed on the trend update weights of each side as follows: The actual control response results are aligned and compared with the evolution trend of key state indicators of wind and fog coupling. The response deviation corresponding to the linkage relationship of each control parameter is calculated. In the collaborative relationship graph, for each edge, the response deviation corresponding to the edge is associated with the trend update weight. When the response deviation is less than the deviation threshold, the trend update weight of the edge is maintained. When the response deviation is greater than the deviation threshold, the trend update weight of the edge is reduced to obtain the deviation correction weight. Based on the deviation correction weights, a structural adaptive adjustment is performed on the synergy relationship graph to generate a structurally adaptive synergy relationship graph. Specifically, the structural adaptive adjustment of the synergy relationship graph involves: Based on the deviation correction weights corresponding to each edge, the parameter associations in the collaborative relationship graph are re-evaluated. When the deviation correction weight of an edge is consistently below the threshold, the effectiveness of the edge in the graph is reduced. When the deviation correction weight of an edge is above the threshold, the association role of the edge in the graph is maintained, and a structurally adaptive collaborative relationship graph is generated. Based on the structurally adaptive cooperative relationship graph, cooperative reorganization and consistency verification under the constraints of the cooperative relationship graph are performed on the initial vectors of the fan control parameters and the initial vectors of the spray control parameters, and the structurally adaptive wind and fog cooperative control quantity is output. The execution of the cooperative reorganization and consistency verification under the constraints of the cooperative relationship graph is as follows: The initial vectors of the fan control parameters and the initial vectors of the spray control parameters are recombined. The parameter value combinations satisfy the parameter association and weight constraints retained in the structural adaptive cooperative relationship graph. Parameter values ​​that do not satisfy the constraints of the structural adaptive cooperative relationship graph are replaced to form candidate parameter combinations. Consistency checks are performed on the candidate parameter combinations to determine whether the fan control parameters and the spray control parameters satisfy the cooperative relationship constraints in terms of adjustment direction, adjustment range, and linkage. When a candidate parameter combination passes the consistency check, it is determined as the structural adaptive wind and mist cooperative control quantity.

[0025] In this embodiment, the output of the actually executable fan control input and spray control input includes: An improved Bouc-Wen model is constructed, which consists of a dual-pathway hysteresis unit, a gating regulation unit, a memory enhancement unit, and a response fusion unit. Specifically, the construction of the improved Bouc-Wen model involves: The original single hysteresis state evolution path in the Bouc-Wen model is split into two parallel hysteresis paths: the fan path and the spray path, resulting in a dual-path hysteresis unit. A state modulation structure gated adjustment unit is introduced based on the hysteresis evolution equation of the Bouc-Wen model. The gated adjustment unit is connected after the dual-path hysteresis unit. The memory enhancement unit is an extended structure introduced based on the Bouc-Wen model relying only on the current hysteresis state. It is connected to the output of the dual-path hysteresis unit. The response fusion unit is a new structural unit added to the output of the Bouc-Wen model, forming an improved Bouc-Wen model. The adaptive wind and mist coordinated control quantity is decomposed into a fan control parameter vector and a spray control parameter vector, and then input into the fan path and spray path of the dual-path hysteresis unit respectively to form a fan hysteresis drive sequence and a spray hysteresis drive sequence. Specifically, the formation of the fan hysteresis drive sequence and the spray hysteresis drive sequence is as follows: According to the order of the control parameters in the time dimension, the value changes of each control parameter vector in the continuous control time are expanded, and according to the timing requirements of the dual-path hysteresis unit for the input sequence, the control parameter vector is time-series rearranged and continuous, so that the fan control parameter vector forms a fan hysteresis drive sequence that changes with time, and the spray control parameter vector forms a spray hysteresis drive sequence that changes with time. The gating control unit generates fan gating and spray gating quantities based on the amplitude and direction of change of the fan hysteresis drive sequence and the spray hysteresis drive sequence, respectively. An asymmetric gating response path is introduced to differentiate the hysteresis evolution weights of the fan path and the spray path, wherein: The generation of fan gate control quantity and spray gate control quantity respectively is specifically as follows: For the fan hysteresis drive sequence and the spray hysteresis drive sequence, the change amplitude and change direction of each drive sequence between adjacent control moments are extracted, the change amplitude is normalized, and the sign of the normalization result is distinguished by the change direction to obtain the intermediate modulation amount. The intermediate modulation amount is mapped to the adjustment coefficient of the corresponding path, and the adjustment coefficient is used as the fan gating amount of the fan path and the spray gating amount of the spray path. Asymmetric gating response path refers to setting independent gating action paths for the fan path and the spray path respectively. The fan gating quantity only acts on the hysteresis evolution process of the fan path, and the spray gating quantity only acts on the hysteresis evolution process of the spray path. Moreover, the two gating action paths are structurally independent and do not share parameters. A segmented hysteresis evolution path is introduced into the dual-path hysteresis unit. Hysteresis evolution processing is performed on the fan hysteresis drive sequence and the spray hysteresis drive sequence based on the fan gating quantity, the spray gating quantity, and the hysteresis evolution weights of the fan path and the spray path. The drive sequence is divided into two response intervals, and the fan hysteresis state variables and spray hysteresis state variables are updated separately in each response interval, resulting in the segmented updated fan hysteresis state variables and spray hysteresis state variables, where: The segmented hysteresis evolution path refers to dividing the hysteresis evolution process into different response intervals on the time axis based on the current control intensity state reflected by the fan gate control quantity and the spray gate control quantity. The hysteresis state is updated in different intervals according to the corresponding evolution path. Each response interval is structurally distinct from each other, but is continuously connected in time, together forming a segmented hysteresis evolution path. The obtained segmented updated fan hysteresis state quantities and spray hysteresis state quantities are specifically as follows: Based on the fan gating quantity, spray gating quantity, and the hysteresis evolution weight of the corresponding path, the fan hysteresis drive sequence and the spray hysteresis drive sequence are divided into intervals in the time dimension. The continuous drive sequence is divided into two response intervals with different control strengths. In each response interval, the hysteresis evolution path corresponding to the response interval is called to update the hysteresis state of the fan path and the spray path every time. When the drive sequence enters the next response interval from one response interval, the hysteresis state quantity at the end of the previous interval is used as the initial state of the new interval to continue to participate in the update, forming the segmented updated fan hysteresis state quantity and spray hysteresis state quantity. The segmented updated fan hysteresis state and spray hysteresis state are input into the memory enhancement unit and introduced into the inertial control sub-channel. The current hysteresis state is combined with the historical hysteresis state to perform inertial adjustment, resulting in the inertial-adjusted fan hysteresis state and the inertial-adjusted spray hysteresis state, where: The inertial control sub-channel refers to an independent state processing path. As a structural channel connecting the historical hysteresis state and the current hysteresis state, it describes the inertial characteristics of the hysteresis state in the continuous evolution of time. The update of the hysteresis state is not only affected by the current driving force, but also by the continuous influence of the past hysteresis evolution results. The inertial adjustment based on the current hysteresis state quantity combined with the historical hysteresis state quantity specifically involves: For each path, the fan hysteresis state quantity and spray hysteresis state quantity obtained from the segmented hysteresis evolution path at the current control time are read. At the same time, the hysteresis state quantity of the corresponding path at the previous control time is extracted from the historical hysteresis state cache. The current hysteresis state quantity and the historical hysteresis state quantity are jointly processed through the inertial control sub-channel. The historical hysteresis state quantity participates in the current state update in proportion, forming an update result that includes the influence of historical evolution, and the fan hysteresis state quantity and spray hysteresis state quantity after inertial adjustment are obtained respectively. The inertial-adjusted fan hysteresis state quantity and the inertial-adjusted spray hysteresis state quantity are input into the response fusion unit. State disturbance suppression filtering is performed on the structurally adaptive wind-mist coordinated control quantity to suppress high-frequency disturbance components. The filtered control quantity is then fused with the corresponding hysteresis state quantity to generate the fan control input and spray control input. The execution of the state disturbance suppression filtering is specifically as follows: The variation characteristics of the wind and fog coordinated control quantity of the structure adaptive system are analyzed along the control time sequence. The control component that exhibits rapid fluctuations between adjacent control times is extracted and identified as a high-frequency disturbance component. By smoothing the high-frequency disturbance component, the overall trend of the control quantity is preserved while the interference caused by instantaneous fluctuations is suppressed, and the filtered wind and fog coordinated control quantity is obtained.

[0026] Example 1: To verify the feasibility of this invention in practice, it was applied to boom spraying operations at a large grain planting base. This area has open terrain, and during spring operations, the natural wind field changes frequently. The wind speed and direction exhibit significant random fluctuations during spraying, easily leading to problems such as spray drift and uneven spray coverage.

[0027] In the operational scenario, the method of this invention is deployed on a self-propelled boom sprayer with a boom width of 24 meters. A continuous-time digital twin model of the spraying operation is constructed to dynamically express the wind-mist coupling relationship between the wind field state, the fan airflow state, and the spray state. During the operation, the credibility of the wind-mist coupling state is continuously modulated and key state indicators are extracted. The evolution trend of the wind-mist state is extrapolated within a short-term operational window. Based on the extrapolation results, the coordinated control quantities of the fan control parameters and the spray control parameters are dynamically generated, and actual control commands are output to achieve coordinated adjustment of the fan airflow and the spray state.

[0028] In field operation tests, the operation time was from 9:00 AM to 4:00 PM, with a single continuous operation area of ​​approximately 120 mu (about 8 hectares). During the operation, the measured surface wind speed fluctuated between 1.2 m / s and 3.6 m / s, with a maximum instantaneous wind direction change angle of approximately 35 degrees. Statistical results show that under complex wind field conditions, the method of this invention can maintain a stable effective spray coverage rate of approximately 91%, an average spray drift distance of approximately 1.9 meters, and good spray deposition uniformity. The operation requires minimal human intervention and there were no interruptions, indicating that the method of this invention has good stability and practicality in complex operating environments.

[0029] Table 1 Performance data of the method of the present invention during field operation

[0030] As shown in Table 1, during a single continuous operation covering approximately 120 acres from 9:00 AM to 4:00 PM, the method of this invention maintained stable operation under various wind conditions. Real-time wind speeds varied between 1.4 m / s and 3.6 m / s during the operation, with a maximum wind direction change angle of 35 degrees, indicating significant uncertainty and dynamism in wind field disturbances. The reliability of the wind-fog coupling state remained generally between 0.82 and 0.89, demonstrating that the method of this invention can reliably reflect the coupling relationship between the wind field, fan airflow, and spray state, providing a reliable state basis for coordinated control.

[0031] From the control behavior data, as the magnitude of wind speed and direction changes increases, the method of this invention can simultaneously increase the fan control adjustment amount and coordinate with the reverse adjustment of the spray parameters. The maximum fan control adjustment amount reaches approximately 14.2%, and the maximum spray parameter adjustment amount reaches approximately -11.3%. The collaborative control response delay is consistently maintained between 0.32 seconds and 0.39 seconds, indicating that the linkage response between the fan and spray control is timely and stable, with no significant control lag. Although the effective spray coverage fluctuates somewhat when wind speed and direction change drastically, it remains stable within the range of approximately 89.9% to 92.1%, and the spray drift distance is controlled between 1.7 meters and 2.1 meters, demonstrating good spray stability and resistance to wind disturbance.

[0032] A comprehensive analysis of the multi-time observation data in Table 1 reveals that the method of this invention can achieve continuous coordinated adjustment of fan airflow and spray state under complex wind field conditions, keeping the spray coverage effect and drift control within an acceptable range. This invention significantly improves the stability and reliability of the operation process while ensuring spray effect, verifying the effectiveness and application value of the method in actual field operations.

[0033] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent collaborative control of wind and mist in a boom sprayer based on digital twins, characterized in that, include: Collect multi-source operation status data of boom sprayers, preprocess the multi-source operation status data, and generate standardized multi-source operation status data; An asynchronous state fusion module based on neural controlled differential equations is constructed to perform continuous-time modeling and state evolution expression of standardized multi-source operation state data with different time scales and sampling frequencies, generating continuous-time digital twin states between wind field state, fan airflow state and spray state. Based on the continuous-time digital twin state, component credibility modulation is performed to generate a wind-fog coupled digital twin state with enhanced credibility. Based on the enhanced credibility of the wind-fog coupled digital twin state, key state indicators of wind-fog coupling are extracted, and short-term state simulation is performed to obtain the evolution trend of wind-fog state during spraying operations, and to form the desired wind-fog coordinated control quantity. Based on the expected wind-fog coordinated control quantity, according to the evolution trend of key state indicators of wind-fog coupling and the actual control response deviation, the coordinated relationship structure between the fan control parameters and the spray control parameters is dynamically adjusted to generate a structurally adaptive wind-fog coordinated control quantity. An improved BoucWen model is constructed to perform response mapping on the structurally adaptive wind and fog coordinated control variables, generate corresponding fan hysteresis state variables and spray hysteresis state variables, and output the fan control input and spray control input that can be actually executed.

2. The intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins as described in claim 1, characterized in that, The multi-source operational status data specifically includes wind farm status data, wind turbine operation status data, sprayer operation status data, boom operation status data, and operational parameter data.

3. The intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins as described in claim 1, characterized in that, The preprocessing of multi-source operational status data specifically includes time consistency processing, abnormal and missing data processing, unit and scale unification processing, noise suppression processing, and data validity labeling.

4. The intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins as described in claim 1, characterized in that, The generated continuous-time digital twin state between the wind field state, the fan airflow state, and the spray state includes: An asynchronous state fusion module based on neural controlled differential equations is constructed. The asynchronous state fusion module consists of a data time processing unit, a continuous time driving unit, a state evolution unit, an observation fusion update unit, and a fusion state output unit. Standardized multi-source operation status data is input into the data time processing unit, and timestamp parsing, time sorting and time alignment are performed on each data component to form an observation sequence arranged in chronological order, and to generate a time interval sequence and a missing measurement marker sequence between adjacent observation times. The observation sequence, time interval sequence, and missing data identification sequence are input into the continuous-time driving unit to construct an adjustable multi-frequency control curve superposition structure. This generates a low-frequency control curve that represents the long-term trend and a high-frequency control curve that represents the short-term disturbance characteristics. These are then weighted and superimposed to form a continuous-time control curve. The continuous-time control curve is input into the state evolution unit, the continuous-time hidden state is set as the internal fusion state, a state update adjustment gate is introduced, and the update amplitude of the continuous-time hidden state is adjusted according to the change characteristics of the continuous-time control curve. A dual-pathway collaborative evolution structure is constructed, and the main path state channel and the offset path state channel are continuously updated in parallel along the time axis under the drive of the neural controlled differential equation to obtain the continuous-time hidden state. At each observation time, the observation vector, missing data identifier sequence and time interval sequence are input into the observation fusion update unit. The observation fusion update unit performs fusion update on the effective observation components under the constraint of the missing data identifier sequence, and adjusts the fusion update amplitude according to the time interval sequence to form the fusion hidden state at the observation time. The fused hidden state is input into the fused state output unit and introduced into the continuous state spectrum decomposition channel. The fused hidden state is subjected to spectrum decomposition processing along the time dimension. The state components corresponding to the wind field state, the fan airflow state and the spray state are frequency-projected and combined respectively to obtain the frequency-aware continuous time state components and generate a continuous time digital twin state.

5. The intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins according to claim 1, characterized in that, The generation of the enhanced credibility wind-fog coupled digital twin state includes: The continuous-time digital twin state is divided into wind field state components, fan airflow state components, and spray state components, forming a component set; Based on the component set, state consistency determination is performed for each state component. The direction and magnitude of continuous change of each state component between adjacent observation times are matched to generate the corresponding consistency deviation. Based on the consistency deviation, physical accessibility is determined for each state component. The accessibility range of the wind field state component, the fan airflow state component, and the spray state component is limited according to the operating parameter data of the boom sprayer, and the accessibility deviation is obtained. Based on the consistency deviation and reachability deviation, component confidence values ​​are generated for each state component. Then, based on the component confidence values, weighted fusion and consistency verification are performed on the wind field state component, the fan airflow state component, and the spray state component to obtain a wind-fog coupled digital twin state with enhanced confidence.

6. The intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins as described in claim 1, characterized in that, The formation of the desired wind and fog coordinated control quantity includes: Based on the enhanced credibility of the wind-fog coupled digital twin state, a wind-fog coupled state sequence consisting of wind field state components, fan airflow state components and spray state components is constructed in chronological order. Based on the wind-fog coupled state sequence, the wind field state component, the fan airflow state component and the spray state component are jointly analyzed to extract key wind-fog coupled state indicators that characterize the wind-fog collaborative operation, forming a set of key wind-fog coupled state indicators. Based on the set of key state indicators of wind and fog coupling, a short-term state simulation window is set, and the time evolution relationship of key state indicators of wind and fog coupling is simulated within the short-term state simulation window to obtain the evolution sequence of key state indicators of wind and fog coupling within the short-term state simulation window. Based on the evolution sequence, a mapping relationship between key state indicators of wind-fog coupling and control parameters is constructed, and the target value range of wind turbine control parameters and spray control parameters is determined. Based on the target value ranges of the fan control parameters and the spray control parameters, the control parameters are matched and conflict resolution is performed to generate the desired wind and fog coordinated control quantity.

7. The intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins according to claim 1, characterized in that, The adaptive wind and fog cooperative control parameters generated by the structure include: The desired wind-fog coordinated control quantity is decomposed into the initial vector of the fan control parameters and the initial vector of the spray control parameters, and a coordinated relationship map is constructed based on the evolution trend of the key state indicators of wind-fog coupling. In the collaborative relationship graph, based on the evolution trend of key state indicators of wind and fog coupling, the initial weights of the coupling strength of each side are updated with trend consistency to obtain the trend update weights, and a collaborative relationship graph reflecting the evolution trend is formed. The actual control response results after executing the fan control input and spray control input are obtained. Based on the deviation between the actual control response results and the evolution trend of the key state indicators of wind and fog coupling, a response deviation amount is generated. Based on the response deviation amount, the trend update weights of each side are updated to perform deviation correction and obtain the deviation correction weight. Based on the deviation correction weights, the cooperative relationship graph is structurally adaptively adjusted to generate a structurally adaptive cooperative relationship graph. Based on the structurally adaptive cooperative relationship graph, the initial vectors of the fan control parameters and the initial vectors of the spray control parameters are subjected to cooperative reorganization and consistency verification under the constraints of the cooperative relationship graph, and the structurally adaptive wind and fog cooperative control quantity is output.

8. The intelligent collaborative control method for wind and mist in a boom sprayer based on digital twins according to claim 1, characterized in that, The outputs can actually execute fan control inputs and spray control inputs, including: An improved Bouc-Wen model is constructed, which consists of a dual-pathway hysteresis unit, a gating regulation unit, a memory enhancement unit, and a response fusion unit. The adaptive wind and fog coordinated control quantity is decomposed into a fan control parameter vector and a spray control parameter vector, and then input into the fan path and spray path of the dual-path hysteresis unit respectively to form a fan hysteresis drive sequence and a spray hysteresis drive sequence. The gating control unit generates fan gating quantity and spray gating quantity respectively based on the change amplitude and change direction of the fan hysteresis drive sequence and the spray hysteresis drive sequence, introduces an asymmetric gating response path, and differentially adjusts the hysteresis evolution weight of the fan path and the spray path. In the dual-path hysteresis unit, a segmented hysteresis evolution path is introduced. Based on the hysteresis evolution weights of the fan gating quantity, the spray gating quantity, and the fan path and spray path, hysteresis evolution processing is performed on the fan hysteresis drive sequence and the spray hysteresis drive sequence. The drive sequence is divided into two response intervals, and the fan hysteresis state quantity and the spray hysteresis state quantity are updated in each response interval to obtain the segmented updated fan hysteresis state quantity and spray hysteresis state quantity. The updated fan hysteresis state and spray hysteresis state are input into the memory enhancement unit and introduced into the inertial control sub-channel. The current hysteresis state is combined with the historical hysteresis state to perform inertial adjustment, so as to obtain the fan hysteresis state and spray hysteresis state after inertial adjustment. The inertial-adjusted fan hysteresis state quantity and the inertial-adjusted spray hysteresis state quantity are input into the response fusion unit. The state disturbance suppression filtering process is performed on the structure-adaptive wind and fog coordinated control quantity to suppress high-frequency disturbance components. The filtered control quantity is then fused with the corresponding hysteresis state quantity to generate the fan control input and the spray control input.

Images