Agricultural machine hybrid power compensation and cooperative control method and system

Abstract

This invention relates to the field of intelligent control technology for agricultural machinery, and particularly to a method and system for compensation and coordinated control of hybrid power systems in agricultural machinery. By acquiring the pre-aiming data, operating condition data, and running status data corresponding to the hybrid agricultural machinery, feedforward compensation and noise filtering are performed by an adaptive state observer with a pre-aiming feedforward input channel to generate a state estimate for the current control cycle. Based on the state estimate and pre-aiming data, the total required torque to maintain the target operating quality and target travel speed is calculated, and power distribution is performed in conjunction with the current operating mode to generate a feedforward power distribution benchmark. Using the feedforward power distribution benchmark as the initial solution, rolling optimization is performed under preset constraints using a model predictive control algorithm to obtain the target torque command for the engine and motor. This achieves efficient coordinated operation of the engine and motor under complex operating conditions, improving the stability of operating quality, the smoothness of driving control, and the energy efficiency of the power system.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for agricultural machinery, and in particular to a method and system for compensation and coordinated control of hybrid power in agricultural machinery. Background Technology

[0002] Currently, with the rapid development of large-scale and intensive modern agriculture, agricultural production has placed higher demands on the operating efficiency, fuel economy, and environmental performance of agricultural machinery. This necessitates upgrading and optimizing agricultural machinery power systems to adapt to complex farmland operation scenarios and alleviate the industry pain points of high energy consumption and high emissions associated with traditional agricultural machinery. However, due to the inherent defects of traditional single-diesel-powered agricultural machinery—high fuel consumption, significant emissions, and poor adaptability to complex operating conditions—hybrid agricultural machinery systems are increasingly adopted. These systems effectively balance the needs of high-torque field operations with energy conservation and emission reduction goals through the complementary use of multiple power sources.

[0003] Existing hybrid agricultural machinery systems are mostly used in complex field operation scenarios. In order to improve the system's operating efficiency and operational stability, preset rules or instantaneous optimization energy management strategies are usually adopted to distribute and adjust the power output of the engine and motor. However, the control strategies of most hybrid agricultural machinery cannot achieve global optimal coordination of multiple power sources according to dynamic operating conditions. It is difficult to force the engine and motor to always operate in their respective high-efficiency ranges. At the same time, the strong coupling characteristics of the driving power and working power of the agricultural machinery are ignored, and there is a lack of advanced compensation capability for predictable operating condition disturbances. There are problems such as control lag and poor adaptability to operating conditions, which leads to high fuel consumption of agricultural machinery. It is impossible to simultaneously take into account the operation quality, operating economy and operation smoothness in complex operating scenarios such as slopes and variable resistance, and it is difficult to give full play to the optimal efficiency of the hybrid system. Summary of the Invention

[0004] This invention provides a method and system for compensation and coordinated control of hybrid power in agricultural machinery, which solves the technical problem that existing hybrid power control technologies for agricultural machinery lack the ability to adapt to complex farmland operation scenarios in terms of power compensation and coordinated control.

[0005] The first aspect of this invention provides a method for compensation and coordinated control of hybrid power in agricultural machinery, applied to hybrid agricultural machinery, the method comprising:

[0006] Acquire the pre-aiming data, working condition data, and operating status data corresponding to the hybrid agricultural machinery;

[0007] An adaptive state observer with a pre-aiming feedforward input channel performs feedforward compensation and noise filtering on the pre-aiming data, the operating condition data, and the running state data to generate a state estimate for the current control cycle.

[0008] Based on the state estimate and the pre-aiming data, the total required torque to maintain the target operation quality and target travel speed is calculated, and the total required torque is distributed according to the current operating mode to generate a feedforward power distribution benchmark.

[0009] Using the feedforward power distribution benchmark as the initial solution, rolling optimization is performed through a model predictive control algorithm under preset constraints to obtain the target torque commands for the engine and motor in the hybrid agricultural machinery.

[0010] Optionally, the step of generating a state estimate for the current control cycle by performing feedforward compensation and noise filtering on the pre-aiming data, the operating condition data, and the running state data using an adaptive state observer with a pre-aiming feedforward input channel includes:

[0011] Based on the longitudinal dynamics, rotational dynamics, and transmission relationship of agricultural machinery, a nonlinear dynamic function matching the hybrid agricultural machinery is constructed, and an adaptive state observer with a pre-aiming feedforward input channel is built by combining the extended Kalman filter algorithm framework.

[0012] The pre-aiming data, the operating condition data, and the running status data are standardized to generate system status data and sensor measurement vectors that are adapted to the adaptive state observer.

[0013] The system state data, the future road condition disturbances in the forecast data, the control command of the previous cycle in the operating condition data, and the state estimate of the previous cycle are input into the forecast feedforward input channel of the adaptive state observer to perform state advance prediction and generate system state prediction values.

[0014] The adaptive state observer performs prediction uncertainty quantification and adaptive weight calculation based on the system state prediction value, the Jacobian matrix of the nonlinear dynamic function, the error covariance matrix of the previous period, the preset process noise covariance matrix, and the preset observation noise covariance matrix, and generates a Kalman gain that matches the current operating condition.

[0015] Based on the Kalman gain, the system state prediction value, and the sensor measurement vector, deviation correction and iterative noise filtering are performed to generate the state estimate value for the current control cycle.

[0016] Optionally, the step of generating a Kalman gain matching the current operating condition by quantifying prediction uncertainty and calculating adaptive weights using the adaptive state observer, based on the predicted system state value, the Jacobian matrix of the nonlinear dynamic function, the error covariance matrix of the previous period, the preset process noise covariance matrix, and the preset observation noise covariance matrix, includes:

[0017] Extract the Jacobian matrix of the nonlinear dynamic function at the system state prediction value with pre-aiming feedforward compensation;

[0018] Using the Jacobian matrix, the error covariance matrix of the previous period, and the preset process noise covariance matrix, the prediction state error covariance matrix with pre-aiming disturbance compensation at the current time is calculated.

[0019] Based on the predicted state error covariance matrix, the observation matrix, and the preset observation noise covariance matrix, an adaptive weighting coefficient is calculated to balance the confidence of the aiming prediction and the confidence of the sensor measurement, thereby generating a Kalman gain that matches the current operating conditions.

[0020] Optionally, the step of performing deviation correction and iterative noise filtering based on the Kalman gain, the system state prediction value, and the sensor measurement vector to generate the state estimate value for the current control cycle includes:

[0021] Using the sensor measurement vector and the system state prediction value, the measurement residual between the sensor measured value and the system state prediction value is calculated to obtain the measurement residual deviation at the current moment;

[0022] The measurement residual bias is weighted using the Kalman gain to obtain the state correction increment at the current moment;

[0023] The system state prediction value is corrected for deviation using the state correction increment to generate an intermediate state correction result.

[0024] The intermediate state correction result is iteratively filtered to suppress the observation noise introduced by the sensor measurement vector and generate the state estimate value within the current control cycle.

[0025] Optionally, the step of calculating the total required torque to maintain the target operating quality and target travel speed based on the state estimate and the pre-aiming data, and distributing the total required torque according to the current operating mode to generate a feedforward power distribution benchmark, includes:

[0026] With the goal of maintaining the target operation quality and target travel speed, a feedforward control model is determined based on the longitudinal dynamics model of agricultural machinery and the load model of the operating implements;

[0027] The feedforward control model performs forward calculations on the state estimate, the future path information and road slope sequence in the forward data within a preset prediction time domain to generate the total demand torque corresponding to each moment in the prediction time domain.

[0028] Based on the principle that both the engine and motor in the hybrid agricultural machinery operate within their respective high-efficiency working ranges, a torque distribution rule adapted to the current working condition is determined.

[0029] According to the torque distribution rule, the total required torque is distributed to obtain the engine feedforward torque reference and the motor feedforward torque reference, and a feedforward power distribution reference is generated.

[0030] Optionally, the step of allocating torque based on a torque distribution rule adapted to the current operating mode, with the principle that both the engine and motor in the hybrid agricultural machinery operate within their respective high-efficiency operating ranges, includes:

[0031] Based on the universal characteristic curve corresponding to the universal characteristic data of the engine in the operating condition data, the optimal fuel consumption rate range is calibrated, and the threshold of the engine's high-efficiency operating range is generated.

[0032] Based on the rated operating parameters of the motor in the operating condition data, the rated high-efficiency operating range is calibrated, and the threshold of the high-efficiency operating range of the motor is generated.

[0033] Using the engine's high-efficiency operating range threshold and the motor's high-efficiency operating range threshold, combined with the hybrid power system characteristic data, agronomic operation demand data, and agricultural machinery industry engineering practice data in the operating condition data, a core principle for torque distribution is formulated, generating a torque distribution principle with the engine and motor in the hybrid agricultural machinery operating in their respective high-efficiency operating ranges as the core.

[0034] Perform operating condition feature identification on the current operating condition data to generate load characteristic parameters and industry standard requirement parameters corresponding to the current operating condition.

[0035] Based on the torque distribution principle, the load characteristic parameters, the industry standard requirement parameters, the engine high-efficiency operating range threshold, and the motor high-efficiency operating range threshold, a torque distribution rule adapted to the current operating condition is constructed.

[0036] Optionally, the step of constructing a torque distribution rule adapted to the current operating condition based on the torque distribution principle, the load characteristic parameters, the industry standard requirement parameters, the engine's high-efficiency operating range threshold, and the motor's high-efficiency operating range threshold includes:

[0037] Based on the load characteristic parameters and the industry standard requirement parameters, an initial candidate rule adapted to the current operating mode is constructed.

[0038] Based on the engine's high-efficiency operating range threshold and the motor's high-efficiency operating range threshold, the initial candidate rules are checked for adaptability, rules that do not meet the preset requirements for efficient operation of dual power sources are eliminated, and target candidate rules are generated.

[0039] Based on the torque distribution principle, the target candidate rules are fine-tuned and optimized to generate torque distribution rules that are adapted to the current operating mode.

[0040] Optionally, the step of using the feedforward power distribution benchmark as the initial solution and performing rolling optimization through a model predictive control algorithm under preset constraints to obtain the target torque command for the engine and motor in the hybrid agricultural machinery includes:

[0041] A multi-objective optimization function is constructed with the goal of minimizing the comprehensive optimization control cost within the preset prediction time domain;

[0042] Using the feedforward power allocation benchmark as the initial value of the control sequence to be optimized, the initial solution for rolling optimization is determined;

[0043] Pre-defined constraint conditions are constructed by employing system dynamics constraints, control input constraints, and dynamic coordination strategy constraints.

[0044] Under the preset constraints, based on the multi-objective optimization function and the initial solution, rolling optimization is performed by the model predictive control algorithm to solve the finite-time domain optimal control problem within the preset prediction time domain, thereby obtaining the target control sequence within the preset control time domain.

[0045] The first control element of the target control sequence is taken as the target torque command for the engine and motor of the hybrid agricultural machinery in the current control cycle.

[0046] Optionally, the dynamic coordination strategy constraints include a motor-dominated mode, an engine-dominated mode, and a hybrid drive mode that match the current operating condition mode;

[0047] The constraints of the motor-dominated mode include that the torque borne by the motor is not less than a preset proportion of the total required torque, the engine working range is limited to a range of zero torque or not less than a preset lower limit of high-efficiency power generation torque and not more than a preset upper limit of high-efficiency power generation torque, the remaining battery power is not less than a preset low power threshold, and the motor torque response rate is not less than a preset rate corresponding to the consistency requirements of agronomic operations.

[0048] The constraints of the engine-dominated mode include: the effective output power of the engine is not less than a preset multiple of the output power of the motor; the load borne by the engine is not less than the basic load that is positively correlated with the load intensity; the rate of change of engine torque does not exceed a preset safety threshold; and the motor only performs limited regenerative braking within the range where the battery charge is not less than a preset allowable threshold.

[0049] The constraints of the hybrid drive mode include that the proportion of motor power to total output power is not lower than a preset lower limit and not higher than a preset upper limit, and the engine working torque is not lower than the lower limit of the high-efficiency range corresponding to the optimal fuel consumption rate and not higher than the upper limit of the high-efficiency range.

[0050] The second aspect of this invention provides a hybrid power compensation and cooperative control system for agricultural machinery, applied to hybrid agricultural machinery, the system comprising:

[0051] The multi-source data acquisition module is used to acquire the pre-aiming data, working condition data and operating status data corresponding to the hybrid agricultural machinery;

[0052] The preview feedforward state estimation module is used to perform feedforward compensation and noise filtering on the preview data, the operating condition data and the running state data through an adaptive state observer with a preview feedforward input channel, and generate the state estimation value of the current control cycle.

[0053] The working condition adaptation feedforward allocation module is used to calculate the total required torque to maintain the target operation quality and target travel speed based on the state estimate and the pre-aiming data, and to allocate the total required torque according to the current working condition mode to generate a feedforward power allocation benchmark.

[0054] The multi-objective rolling optimization module is used to perform rolling optimization using the feedforward power distribution benchmark as the initial solution and under preset constraints through a model predictive control algorithm to obtain the target torque commands of the engine and motor in the hybrid agricultural machinery.

[0055] As can be seen from the above technical solutions, the present invention has the following advantages:

[0056] This invention addresses the technical problem of existing hybrid power control technologies for agricultural machinery lacking the ability to adapt to complex farmland operation scenarios in terms of power compensation and coordinated control. It acquires the pre-aiming data, operating condition data, and running status data corresponding to the hybrid agricultural machinery. An adaptive state observer with a pre-aiming feedforward input channel performs feedforward compensation and noise filtering on the aforementioned data to generate a state estimate for the current control cycle. Then, based on this state estimate and the pre-aiming data, it calculates the total torque required to maintain the target operation quality and target travel speed, and combines this with the current operating mode to generate a feedforward power allocation benchmark. Finally, the feedforward power allocation benchmark is used as the initial... Solution: Under preset constraints, the target torque command of the engine and motor is obtained through rolling optimization using a model predictive control algorithm. Through the above steps, the predictable disturbances are addressed by utilizing the advance compensation capability of feedforward control, and dynamic optimal coordination under multiple constraints is achieved through multi-objective rolling optimization. At the same time, the dynamic coordination strategy forces the engine and motor to operate in their respective high-efficiency ranges, and global optimal energy efficiency management of the power source is achieved by optimizing instantaneous fuel consumption, avoiding inefficient engine operation or ineffective power consumption of the motor. Thus, in complex farmland operation scenarios, the overall improvement of operation quality, economy and smoothness is achieved, effectively adapting to the control requirements of complex operation conditions. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 A flowchart illustrating the steps of a hybrid power compensation and coordinated control method for agricultural machinery provided in an embodiment of the present invention;

[0059] Figure 2 This is a schematic diagram of a hybrid intelligent agricultural machine that consists of a dual power source, an engine and an electric motor, and is required to drive independent end-effectors.

[0060] Figure 3 This is a schematic diagram of the structure of a hybrid intelligent agricultural machinery hybrid power compensation and cooperative control system under complex working conditions, provided by an embodiment of the present invention;

[0061] Figure 4 This is a structural block diagram of a hybrid power compensation and cooperative control system for agricultural machinery provided in an embodiment of the present invention. Detailed Implementation

[0062] This invention provides a method and system for compensation and coordinated control of hybrid power in agricultural machinery, which addresses the technical problem that existing hybrid power control technologies for agricultural machinery lack the ability to adapt to complex farmland operation scenarios in terms of power compensation and coordinated control.

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should be noted that in the optional embodiments of the present invention, the object information and other related data involved require the permission or consent of the object when the embodiments of the present invention are applied to specific products or technologies, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. That is to say, if the embodiments of the present invention involve data related to the object, it needs to be obtained with the authorization and consent of the object, the authorization and consent of the relevant departments, and in compliance with the relevant laws, regulations, and standards of the country and region. If personal information is involved in the embodiments, the acquisition of all personal information requires the consent of the individual. If sensitive information is involved, the separate consent of the information subject is required, and the embodiments also need to be implemented with the authorization and consent of the object.

[0064] Example 1

[0065] Please see Figure 1 , Figure 1 A flowchart illustrating the steps of a hybrid power compensation and coordinated control method for agricultural machinery provided in an embodiment of the present invention.

[0066] This invention provides a method for compensation and coordinated control of hybrid power in agricultural machinery, applicable to hybrid agricultural machinery. Hybrid agricultural machinery refers to hybrid agricultural machinery that consists of a dual power source (engine and motor) and needs to drive independent end implements. For example... Figure 2 The diagram illustrates a hybrid intelligent agricultural machine with a dual power source consisting of an engine and a drive motor, capable of driving independent end-effectors. This hybrid intelligent agricultural machine employs a dual power source architecture consisting of an engine and a drive motor, coupled with an independent end-effector drive chain. The core power and mechanical execution architecture is divided into two independent mechanical chains: a walking drive and an operating drive. The fuel engine serves as the primary power source for field operations, mechanically driving a generator and an electromagnetic clutch. The electromagnetic clutch, through mechanical connections, sequentially connects to the operating reducer and the power output shaft. The power output shaft can flexibly switch between three types of end-effectors—rotary tillage, plant protection, and weeding—achieving independent drive for field operations. The engine is equipped with a dedicated engine electronic control unit for precise control of operating conditions. The drive motor serves as the primary power source for walking, mechanically connecting to the walking reducer and wheels to form a complete walking power chain. A dedicated drive motor controller enables speed regulation and operating condition control.

[0067] This hybrid intelligent agricultural machine uses a vehicle controller as the top-level collaborative control hub to construct a closed-loop electronic control architecture for the entire system. The vehicle controller connects to the engine electronic control unit, battery management system, drive motor controller, and electromagnetic clutch via control links. Specifically, the engine electronic control unit enables unified regulation of engine start / stop and output power; the battery management system enables status monitoring and charge / discharge management of the energy storage unit; the drive motor controller enables precise speed adjustment of walking power; and the electromagnetic clutch enables on / off control of working power, thus achieving coordinated matching of the two power sources and linked regulation of walking and working functions.

[0068] This hybrid intelligent agricultural machine is equipped with a tiered power supply and energy storage system, consisting of a high-voltage main energy storage unit and a low-voltage auxiliary power supply unit, providing a stable energy supply for all electrical components of the vehicle. The core of the high-voltage main energy storage unit is a 48V battery, which connects to the battery management system, drive motor, and generator via power supply connections. The generator, driven by the engine, can replenish the 48V battery via the power supply connection. The battery management system manages the charging and discharging of the 48V battery throughout its entire lifecycle via the power supply connection. The low-voltage auxiliary power supply unit includes a 12V battery and a 24V power supply. The 12V battery provides power to the engine electronic control unit via the power supply connection, while the 24V power supply provides power to the electromagnetic clutch via the power supply connection. The 12V battery and 24V power supply form a complementary link via the power supply connection, ensuring the stable operation of the low-voltage electronic control components.

[0069] The methods include:

[0070] Step 101: Obtain the pre-aiming data, working condition data, and operating status data corresponding to the hybrid agricultural machinery.

[0071] In this embodiment of the invention, various sensors mounted on intelligent agricultural machinery, i.e., hybrid agricultural machinery, acquire in real time the pre-aiming data, operating condition data, and operational status data required by the system. Specifically: Pre-aiming data consists of the road slope sequence and road surface roughness characteristics within a preset distance ahead of the agricultural machinery, obtained through forward sensing sensors (such as vision sensors, lidar, or millimeter-wave radar). This data is used to detect future road condition disturbances in advance, providing core input for pre-aiming feedforward compensation. Operating condition data refers to the operation scenario mode signal set by the operator, including at least three typical agricultural machinery operation modes: rotary tillage, branch shredding, and weeding. This data is used to distinguish different operation load characteristics and control objectives, providing an operating condition benchmark for subsequent power distribution strategies. Operational status data includes the longitudinal speed of the agricultural machinery, engine speed, motor speed, and the raw load torque signal measured by a torque sensor installed on the output shaft at the end of the transmission mechanism. This data characterizes the current real-time operating status of the agricultural machinery, providing basic measurement input for the state observer. Through multi-source sensor collaborative acquisition, the pre-aiming data, operating condition data, and operational status data are acquired synchronously, providing complete and reliable input support for subsequent adaptive state observation and hybrid power collaborative control.

[0072] Step 102: Use an adaptive state observer with a preview feedforward input channel to perform feedforward compensation and noise filtering on the preview data, operating condition data and running state data to generate the state estimate value for the current control cycle.

[0073] In this embodiment of the invention, pre-aiming data, operating condition data, and noisy operating status data are input to an adaptive state observer with a pre-aiming feedforward input channel. The adaptive state observer uses the future slope disturbance in the pre-aiming data as a known input in a feedforward manner, and uses the work load characteristics corresponding to the operating condition data as a priori reference for state prediction. A nonlinear dynamic model is constructed based on the longitudinal dynamics, rotational dynamics, and transmission relationship of the agricultural machinery. An extended Kalman filter algorithm is used to achieve adaptive state estimation. Sensor measurement noise in the operating status data is suppressed through algorithm iteration, and finally, a high-confidence state estimate for the current control cycle is output. The state estimate includes four core states: actual agricultural machinery speed, actual end-load torque, engine speed, and motor speed. Simultaneously, the comprehensive driving resistance currently experienced by the agricultural machinery can be derived from these core states, providing a reliable state basis for subsequent calculation of the total demand torque, power distribution, and coordinated control of the hybrid power system. The longitudinal dynamics of the agricultural machinery can also be the longitudinal dynamics of the vehicle.

[0074] Further, step S102 includes the following steps:

[0075] S11. Based on the longitudinal dynamics, rotational dynamics, and transmission relationship of agricultural machinery, a nonlinear dynamic function matching hybrid agricultural machinery is constructed, and an adaptive state observer with a pre-aiming feedforward input channel is built by combining the extended Kalman filter algorithm framework.

[0076] In the embodiments of the present invention, it is first clarified that... System state at time 1 Defined as:

[0077] ;

[0078] in, This refers to the longitudinal speed of the agricultural machinery. This refers to the output shaft load torque. and These refer to the engine and motor speeds, respectively.

[0079] Based on this, and according to the longitudinal dynamics, rotational dynamics, and transmission system characteristics of agricultural machinery, a nonlinear dynamic function describing the state change law of the system is established:

[0080] ;

[0081] in, This represents the current system status, including system status data such as the longitudinal speed of the agricultural machinery, the load torque of the end-effectors, the engine speed, and the motor speed. The system state at the previous moment has the same dimension as the system state vector at the current moment. The control input vector is composed of the torque commands sent to the engine and motor in the previous moment; Let be the future road condition disturbance vector obtained from the prediction, where To obtain the road slope angle in advance, This is the road surface resistance coefficient. It is process noise, representing model uncertainty; This is a nonlinear mapping relationship established based on the dynamic characteristics of agricultural machinery, namely, the longitudinal dynamics, rotational dynamics, and transmission system characteristics of the vehicle.

[0082] Based on this nonlinear dynamic function, and combined with the extended Kalman filter algorithm framework, an adaptive state observer with a pre-aiming feedforward input channel is built. The pre-aiming disturbance is introduced into the filtering process as a feedforward quantity, which improves the adaptability and anticipatory response capability of the state prediction to road condition changes.

[0083] S12. Standardize the preview data, operating condition data, and running status data to generate system status data and sensor measurement vectors that are adapted to the adaptive state observer.

[0084] In this embodiment of the invention, the pre-aiming data, working condition data, and operating status data are uniformly standardized and converted into system status data and sensor measurement vectors that can be directly used by the observer. The system status data consists of the longitudinal speed of the agricultural machinery, the torque of the working load, the engine speed, and the motor speed, forming a four-dimensional state vector. The sensor measurement vector consists of observable signals collected by the onboard sensors, satisfying the observation relationship, i.e., the state observation equation:

[0085] ;

[0086] in, For sensor measurement vectors (such as vehicle speed from wheel speed sensors, measurement values ​​from torque sensors, etc.); The observation matrix; for The system state at a given moment is the system state data at the current moment. To reduce observation noise, standardization is used to ensure that multi-source data maintains consistency in units and numerical range, guaranteeing stable observation calculations and reliable output.

[0087] S13. Input the system state data, the future road condition disturbances in the preview data, the control command of the previous cycle in the operating condition data, and the state estimate of the previous cycle into the preview feedforward input channel of the adaptive state observer to perform state advance prediction and generate the system state prediction value.

[0088] In this embodiment of the invention, the system state data, future road condition disturbances in the preview data, the control command of the previous cycle in the operating condition data, and the state estimate of the previous cycle are input into the preview feedforward channel of the adaptive state observer to perform state advance prediction and obtain the system state prediction value:

[0089] ;

[0090] in, This is the predicted system state value for the current control cycle, i.e., the current state predicted based on information from the previous cycle. is the system state estimate of the previous control cycle, and is the optimal state value after deviation correction and noise filtering; This is the control input vector for the previous control cycle, consistent with the definition of the control input vector in the aforementioned nonlinear dynamics function; The vector for predicting future road condition disturbances obtained in the previous control cycle includes the road slope angle and road surface resistance coefficient, consistent with the aforementioned definition; This is a nonlinear mapping relationship established based on the dynamic characteristics of agricultural machinery, namely, the longitudinal dynamics, rotational dynamics, and transmission system characteristics of the vehicle.

[0091] By using the preview data as a feedforward input to participate in state prediction, the prediction results can reflect the impact of future road conditions on the state of agricultural machinery in advance, thereby improving prediction accuracy and proactive response capability.

[0092] S14. Using an adaptive state observer, based on the system state prediction value, the Jacobian matrix of the nonlinear dynamic function, the error covariance matrix of the previous cycle, the preset process noise covariance matrix, and the preset observation noise covariance matrix, perform prediction uncertainty quantification and adaptive weight calculation to generate a Kalman gain that matches the current operating condition.

[0093] In this embodiment of the invention, the adaptive state observer completes the prediction uncertainty quantification and adaptive weight calculation based on the system state prediction value, Jacobian matrix, previous cycle error covariance matrix, preset process noise covariance matrix and preset observation noise covariance matrix, and finally outputs a Kalman gain that adapts to the current operating conditions.

[0094] Furthermore, step S14 includes the following steps:

[0095] S141. Extract the Jacobian matrix of the nonlinear dynamic function at the system state prediction value with anticipation feedforward compensation.

[0096] In this embodiment of the invention, a first-order Taylor expansion is performed on the nonlinear dynamic function at the predicted system state value to obtain the corresponding Jacobian matrix. This Jacobian matrix is ​​used for the recursive calculation of subsequent error covariance. It reflects the local variation of the nonlinear dynamic function at the current state prediction value, providing a linearized model basis for the calculation of the subsequent predicted state error covariance matrix.

[0097] S142. Using the Jacobian matrix, the error covariance matrix of the previous period, and the preset process noise covariance matrix, calculate the prediction state error covariance matrix with pre-aiming disturbance compensation at the current time.

[0098] In this embodiment of the invention, the uncertainty of state prediction is quantified and recursively calculated using the Jacobian matrix, the error covariance matrix of the previous period, and the preset process noise covariance matrix to obtain the predicted state error covariance matrix with pre-aiming perturbation compensation at the current time, expressed as:

[0099] ;

[0100] in, This is the predicted state error covariance matrix for the current control cycle, used to quantify the uncertainty of the system state prediction. This is the Jacobian matrix for the current control cycle, i.e., the local linearization matrix; This is the error covariance matrix of the previous control cycle, used to characterize the uncertainty of the state estimate at the previous time step; Jacobian matrix The transpose of the matrix; The preset process noise covariance matrix is ​​a constant matrix used to characterize model uncertainty, unmodeled dynamics, and errors caused by pre-aiming disturbances. It is preset by experimental calibration.

[0101] S143. Based on the predicted state error covariance matrix, the observation matrix, and the preset observation noise covariance matrix, calculate the adaptive weighting coefficients that balance the confidence of the aiming prediction and the confidence of the sensor measurement, and generate the Kalman gain that matches the current operating conditions.

[0102] In this embodiment of the invention, a Kalman gain is calculated based on the predicted state error covariance matrix, the observation matrix, and the preset observation noise covariance matrix. This enables adaptive allocation of the prediction confidence and sensor measurement confidence. The Kalman gain automatically adjusts its weights according to the operating conditions, increasing the model prediction weights when sensor noise is high and increasing the measurement signal weights when road conditions change abruptly, thus achieving adaptive state estimation. The expression for the Kalman gain is:

[0103] ;

[0104] in, is the Kalman gain for the current control cycle, and is the adaptive weighting coefficient used to balance the reliability of the aiming prediction and the sensor measurement; The predicted state error covariance matrix for the current control cycle is consistent with the matrix obtained by the aforementioned recursive calculation. Observation matrix The transpose of the matrix; The observation matrix is ​​used to map the system state vector into an observable vector with the same dimension as the sensor measurement vector; The preset observation noise covariance matrix is ​​a constant matrix used to characterize the intensity of sensor measurement noise, and is pre-calibrated by the sensor accuracy parameters; For matrix The inverse matrix.

[0105] S15. Based on Kalman gain, system state prediction, and sensor measurement vectors, deviation correction and iterative noise filtering are performed to generate the state estimate for the current control cycle.

[0106] Further, step S15 includes the following steps:

[0107] S151. Using the sensor measurement vector and the system state prediction value, calculate the measurement residual between the sensor measured value and the system state prediction value to obtain the measurement residual deviation at the current moment.

[0108] In this embodiment of the invention, the predicted system state value is mapped to the corresponding predicted measurement value through the observation equation. Then, the difference between the predicted measurement value and the sensor measurement vector (i.e., the measurement data actually collected by the sensor) is calculated to obtain the measurement residual deviation. This deviation directly reflects the degree of deviation between the predicted state and the actual physical state. The magnitude of the deviation represents the order of magnitude of the prediction error, and the positive or negative sign of the deviation represents the direction of deviation of the predicted value from the true value.

[0109] S152. The measurement residual bias is weighted using Kalman gain to obtain the state correction increment at the current moment.

[0110] In this embodiment of the invention, the measurement residual bias is weighted using Kalman gain to obtain the state correction increment. This increment is adaptively adjusted by an adaptive state observer, which can quickly correct the state bias while suppressing noise, ensuring that the correction magnitude is both reasonable and accurate, and avoiding state oscillations caused by overcorrection.

[0111] S153. Use state correction increments to correct the deviation of the system state prediction value and generate intermediate state correction results.

[0112] In this embodiment of the invention, the state correction increment is superimposed on the system state prediction value to complete the deviation correction of the predicted state, obtain the intermediate state correction result, and make the state estimation converge to the true state, providing the basic input for subsequent iterative filtering.

[0113] S154. Iteratively filter the intermediate state correction results to suppress the observation noise introduced by the sensor measurement vector and generate the state estimate value within the current control cycle.

[0114] In this embodiment of the invention, the intermediate state correction result is iteratively filtered and updated to suppress the influence of sensor observation noise and model process noise, ultimately obtaining the state estimate for the current control cycle. This state estimate is characterized by low noise, high accuracy, and fast response, and can truly reflect the real-time operating state of the hybrid agricultural machinery, providing a stable and reliable state input for subsequent torque calculation, power distribution, and model predictive control. The expression corresponding to the state estimate is:

[0115] ;

[0116] in, The system state estimate for the current control cycle, i.e. the optimal state value after deviation correction and iterative filtering, provides the core input for subsequent control. This is the predicted system state value for the current control cycle; The Kalman gain for the current control cycle; The sensor measurement vector at the current moment is composed of measured data collected by on-board sensors such as wheel speed sensors and torque sensors; The observation matrix; To measure residual deviation.

[0117] Step 103: Based on the state estimate and the target aiming data, calculate the total required torque to maintain the target operation quality and target travel speed, and distribute the total required torque according to the current working mode to generate a feedforward power distribution benchmark.

[0118] In this embodiment of the invention, based on high-precision state estimation values ​​and combined with forward-looking data, forward torque calculation and power distribution for working condition adaptation are completed. Finally, a forward power distribution benchmark with clear physical meaning and fast convergence speed is generated as the initial solution for subsequent model predictive control rolling optimization. At the same time, in the initial distribution stage, it is ensured that the engine and motor operate in their respective efficient working ranges as much as possible.

[0119] Further, step S103 includes the following steps:

[0120] S21. To maintain the target operation quality and target travel speed, a feedforward control model is determined based on the longitudinal dynamics model of agricultural machinery and the load model of the operating implements.

[0121] In this embodiment of the invention, maintaining the target operation quality and the target travel speed are the control objectives. Based on a simplified agricultural machinery longitudinal dynamics model and power flow model, a system is built that defines a prediction time domain. The feedforward control model is used to calculate the feedforward quantity in the corresponding time domain. The expression of the feedforward control model is:

[0122] = ;

[0123] in, This is the current control cycle, which is the reference cycle number for this feedforward control calculation and the time origin for all timing calculations. The time index (offset) within the prediction time domain has a value range of [value range missing]. This represents the period relative to the current control cycle. Prediction step size: =0 corresponds to the current control cycle. itself, =1 corresponds to the next control cycle, and so on. = Corresponds to the furthest future control period within the preset prediction time domain; To predict the target time sequence number in the time domain, the current control cycle is used. With predicted offset The combination yields a specific future moment that this calculation is targeting. For the current control cycle Regarding future goals The calculated feedforward control reference is ultimately decomposed into the engine feedforward torque reference and the motor feedforward torque reference. A feedforward nonlinear mapping relationship is established based on the longitudinal dynamics of agricultural machinery, the load characteristics of operating implements, and the torque distribution rules for working conditions. To predict the target time in the time domain The system state variables are derived from the look-ahead extension of the state estimates generated in the preceding steps; For the target time in the pre-aiming data The corresponding road slope sequence; The preset target driving speed; For the target time The baseline load torque determined by the current operating mode; The preset prediction time domain represents the period from the current control cycle. The maximum number of control cycles that can be predicted forward. In the current control cycle Below, regarding future goals and moments The calculated reference value of the feedforward torque allocated to the motor; In the current control cycle Below, regarding future goals and moments The calculated feedforward torque reference value allocated to the engine.

[0124] S22. Using a feedforward control model, perform forward calculations on the state estimate, future path information and road slope sequence in the forecast data within the preset prediction time domain to generate the total demand torque corresponding to each moment in the prediction time domain.

[0125] In this embodiment of the invention, based on the feedforward control model determined in step S21, within a preset prediction time domain, the input state estimate, future path information from the look-ahead data, and road slope sequence are used to perform look-ahead calculations to obtain the total demand torque corresponding to each moment in the prediction time domain:

[0126] ;

[0127] in, This is the current control cycle, which is the reference cycle number for this feedforward control calculation and the time origin for all timing calculations. The time index (offset) within the prediction time domain has a value range of 1. ; To predict the target time sequence number in the time domain, the current control cycle is used. With predicted offset The combination yields a specific future moment that this calculation is targeting. For the target time in the pre-aiming data The corresponding road slope sequence; The preset target driving speed; For the target time The baseline load torque determined by the current operating mode; The preset prediction time domain represents the period from the current control cycle. The maximum number of control cycles that can be predicted forward. For the current control cycle Regarding future goals Calculate the total required torque; The radius of the wheel; , These are the driving and operating gear ratios, respectively; , These are the corresponding efficiencies of the driving and operating transmission systems, respectively. For the overall vehicle weight; It is the acceleration due to gravity; This is the rolling resistance coefficient; air density; This refers to the drag coefficient; This refers to the windward area.

[0128] By combining the predicted road slope sequence in the prediction time domain to complete the forward calculation of total torque demand, the torque demand can be matched in advance with future road conditions and work load changes, providing a precise basic input for subsequent power distribution.

[0129] S23. Based on the principle that both the engine and motor in the hybrid agricultural machinery operate within their respective high-efficiency working ranges, determine the torque distribution rules that are compatible with the current working condition.

[0130] In this embodiment of the invention, the core allocation principle of the hybrid agricultural machinery is that both the engine and the motor operate in their respective high-efficiency working ranges. Combining the load characteristics of the current working mode and the requirements of industry standards, a torque allocation rule adapted to the current working condition is constructed to provide an execution basis for the breakdown of the total required torque. The core selection criterion is to ensure that the engine and the motor operate in their respective high-efficiency ranges as much as possible in the initial allocation stage.

[0131] Furthermore, step SS23 includes the following steps:

[0132] S231. Based on the universal characteristic curve corresponding to the universal characteristic data of the engine in the operating condition data, the optimal fuel consumption rate range is calibrated, and the threshold of the engine's efficient operating range is generated.

[0133] In this embodiment of the invention, based on the universal characteristic curve corresponding to the engine's universal characteristic data in the operating condition data, the optimal fuel consumption rate (BSFC) range is calibrated, and the engine's efficient operating range threshold is generated. The engine's efficient operating range threshold is the upper and lower limits of torque and speed within the operating range corresponding to the engine's optimal fuel consumption rate. This provides an efficient operating verification standard on the engine side for subsequent torque distribution, and the engine preferentially operates within this range to achieve optimal fuel economy.

[0134] S232. Based on the rated operating parameters of the motor in the operating condition data, calibrate the rated high-efficiency operating range and generate the threshold of the high-efficiency operating range of the motor.

[0135] In this embodiment of the invention, based on the rated operating parameters of the motor in the operating condition data, the rated high-efficiency operating range of the motor is calibrated, and a threshold value for the high-efficiency operating range of the motor is generated. The threshold value for the high-efficiency operating range of the motor is the upper and lower limits of torque and speed within the rated high-efficiency operating range of the motor. Together with the threshold value for the high-efficiency operating range of the engine, they constitute a verification standard for the high-efficiency operation of dual power sources. The motor preferentially operates within this range to achieve optimal energy utilization efficiency.

[0136] S233. By adopting the high-efficiency operating range threshold of the engine and the high-efficiency operating range threshold of the motor, and combining the hybrid power system characteristic data, agronomic operation demand data and agricultural machinery industry engineering practice data in the operating condition data, the core principle of torque distribution is formulated, and a torque distribution principle with the engine and motor in hybrid agricultural machinery operating in their respective high-efficiency operating ranges is generated.

[0137] In this embodiment of the invention, based on the high-efficiency operating range thresholds of the engine and the electric motor, and combined with hybrid power system characteristic data, agronomic operation requirement data, and agricultural machinery industry engineering practice data, a torque distribution principle is formulated with the engine and electric motor operating within their respective high-efficiency operating ranges. The agronomic operation requirement data corresponds consistently to the definition of target operation quality; the agricultural machinery industry engineering practice data includes national and industry standards such as GB / T 5262-2008.

[0138] S234. Perform operating condition feature identification on the current operating condition mode data to generate the load characteristic parameters and industry standard requirement parameters corresponding to the current operating condition mode.

[0139] In this embodiment of the invention, the current operating condition mode data is subjected to operating condition feature identification to distinguish the type of the current operating condition, and the load characteristic parameters and industry standard requirement parameters corresponding to the current operating condition mode are extracted and generated. The operating condition modes are divided into four categories:

[0140] (1) Rotary tillage mode: The load characteristic parameters are large and relatively stable load torque, which are directly related to soil hardness and tillage depth, and have high requirements for consistency of working depth; the industry standard requirements parameters are the requirements for tillage depth stability in GB / T 5262-2008, which require the power system to provide continuous and stable large torque;

[0141] (2) Fragmentation mode: The load characteristic parameters are that the load has strong impact and periodicity, high peak torque, and high requirements for the instantaneous overload capacity and dynamic response of the system; the industry standard requirement parameters are the industry's requirements for operating efficiency and equipment reliability, and it is necessary to avoid the impact load from causing the engine to stall or the transmission system to overload.

[0142] (3) Weeding / Light Load Medium and High Speed ​​Operation Mode: The load characteristic parameter is that the load torque is relatively small and the uniformity of the operation speed is required; the industry standard requirement parameter is that the agronomic requirements are that the stubble is neat and there are no missed cuts, and the chassis speed is highly stable.

[0143] (4) Ramp transport / transfer mode: The load characteristic parameters are mainly to overcome the slope resistance, the driving power demand increases significantly, and the operating equipment may be unloaded or lightly loaded; the industry standard requirements parameters are safety and ramp passability requirements, and sufficient traction reserve and stable speed control are required.

[0144] S235. Based on torque distribution principles, load characteristic parameters, industry standard requirements parameters, engine high-efficiency operating range thresholds, and motor high-efficiency operating range thresholds, a torque distribution rule adapted to the current operating condition is constructed.

[0145] In this embodiment of the invention, based on torque distribution principles, load characteristic parameters, industry standard requirement parameters, engine high-efficiency operating range thresholds, and motor high-efficiency operating range thresholds, a three-step screening method is used to construct torque distribution rules that are compatible with the current operating mode. The rules are divided into three categories: engine-dominated mode, motor-dominated mode, and hybrid drive mode.

[0146] Further, step S235 includes the following steps:

[0147] S2351. Based on load characteristic parameters and industry standard requirement parameters, construct initial candidate rules that are adapted to the current operating mode.

[0148] In this embodiment of the invention, based on the load characteristic parameters and industry standard requirement parameters corresponding to the current working condition, initial candidate rules adapted to the current working condition mode are determined, such as rotary tillage mode and branch shredding mode, and initial candidate rules of engine-dominated type are initially screened; for weeding / light load medium and high speed operation mode, initial candidate rules of hybrid drive type are initially screened; for slope transportation / transfer mode, initial candidate rules of motor-dominated type are initially screened.

[0149] S2352. Based on the high-efficiency operating range thresholds of the engine and the high-efficiency operating range thresholds of the motor, the initial candidate rules are checked for adaptability, rules that do not meet the preset requirements for high-efficiency operation of dual power sources are eliminated, and target candidate rules are generated.

[0150] In this embodiment of the invention, the high-efficiency operating range threshold of the engine and the high-efficiency operating range threshold of the motor are used as verification standards to perform dual-power-source high-efficiency operation adaptability verification on the initial candidate rules, eliminate rules that do not meet the preset dual-power-source high-efficiency operation requirements, retain the rules that meet the requirements, and generate target candidate rules.

[0151] S2353. Based on the torque distribution principle, the target candidate rules are fine-tuned and optimized to generate torque distribution rules that are compatible with the current working condition.

[0152] In this embodiment of the invention, the core principle of torque distribution for efficient operation of dual power sources is taken into account. Combined with the current real-time battery remaining charge (SOC) status, load fluctuations, and road condition information in the pre-targeting data, the target candidate rules are fine-tuned and optimized to generate a final torque distribution rule adapted to the current operating mode. The core contents of the three types of rules are as follows:

[0153] (1) Engine-dominant mode rule: The engine output accounts for 70%~90% of the total required torque, and the motor is responsible for the remaining part and instantaneous fluctuation compensation (10%~30%). When the load suddenly increases, the motor responds quickly. The engine torque set point is determined by looking up the table according to the load torque and prioritizes operation in the optimal fuel consumption rate range.

[0154] (2) Motor-dominated mode rule: The motor bears 60%~80% of the dynamic load. The engine only intervenes when the battery power is low and the power demand exceeds the motor capacity, or runs in power generation mode to charge the battery. The motor torque is controlled by the vehicle speed closed loop to ensure smooth driving and avoid low-load and inefficient engine operation.

[0155] (3) Hybrid drive mode rule: The engine and motor output in a coordinated ratio of 50%~70%:30%~50%. The specific ratio is dynamically adjusted by the slope and battery SOC. The engine provides continuous high torque, and the motor provides peak compensation and dynamic adjustment. The engine operates in the high torque and high efficiency zone, and the motor dynamically compensates according to the slope to ensure the optimal overall system efficiency.

[0156] S24. Distribute the total required torque according to the torque distribution rules, decompose it to obtain the engine feedforward torque reference and the motor feedforward torque reference, and generate the feedforward power distribution reference.

[0157] In this embodiment of the invention, based on a reference allocation rule, the total demand torque is decomposed into feedforward references for the engine and the motor. According to the torque allocation rule finally determined in step S23, the total demand torque at each moment in the prediction time domain is allocated using dual power sources, decomposing to obtain the engine feedforward torque reference and the motor feedforward torque reference at the corresponding moment, and finally generating the feedforward power allocation reference:

[0158] (1) Expression of the feedforward reference vector at a single time step:

[0159] ;

[0160] (2) Complete feedforward control sequence expression:

[0161] ;

[0162] in, This is the current control cycle, which is the reference cycle number for this feedforward control calculation and the time origin for all timing calculations. The time index (offset) within the prediction time domain has a value range of [value range missing]. ; To predict the target time sequence number in the time domain, the current control cycle is used. With predicted offset The combination yields a specific future moment that this calculation is targeting. For the current control cycle Regarding future goals The calculated feedforward control reference is ultimately decomposed into the engine feedforward torque reference and the motor feedforward torque reference. The preset prediction time domain represents the period from the current control cycle. The maximum number of control cycles that can be predicted forward. For the current control cycle Regarding future goals Calculated engine feedforward torque reference; For the current control cycle Regarding future goals The calculated motor feedforward torque reference; For length is The complete feedforward control sequence, i.e. the final generated feedforward dynamics allocation reference, serves as a set of initial solutions for subsequent model predictive control.

[0163] Step 104: Using the feedforward power distribution benchmark as the initial solution, roll optimization is performed using the model predictive control algorithm under preset constraints to obtain the target torque commands for the engine and motor in the hybrid agricultural machinery.

[0164] In this embodiment of the invention, the feedforward power distribution benchmark generated in step S103 is used as the initial solution. Under the constraints of system dynamics, control input, and dynamic coordination strategy, a multi-objective optimization function is constructed with the core of operation quality stability, vehicle speed tracking accuracy, power switching smoothness, and system instantaneous energy efficiency. The model predictive control (MPC) algorithm is used to solve the finite-time optimal control problem online in a rolling manner. Finally, the globally optimal engine target torque command and motor target torque command within the current control cycle are output.

[0165] Further, step S104 includes the following steps:

[0166] S31. Construct a multi-objective optimization function with the goal of minimizing the comprehensive optimization control cost within the preset prediction time domain.

[0167] In this embodiment of the invention, the core of model predictive control is to achieve multi-demand coordination through quantified optimization objectives. Therefore, the core demands of agricultural machinery operation (stable operation quality, precise speed tracking) and system operation demands (smooth power switching, optimal fuel efficiency) need to be transformed into quantifiable and solvable mathematical functions, i.e., multi-objective optimization functions. By setting the minimum comprehensive optimization control cost as the core objective, the four-dimensional demands are decomposed into independent optimization sub-items. Then, the priority of each sub-item is balanced through a weight matrix to ensure that the optimization result meets both agronomic operation requirements and system operating efficiency and smoothness. Its core function is to provide a clear objective guide for subsequent rolling optimization, avoiding a decline in operational performance or energy efficiency caused by single-objective optimization. Therefore, using a model predictive control framework, the power compensation and coordination problem is expressed as a constrained rolling optimization problem, with each time step... The multi-objective optimization function constructed to solve the finite-time optimal control problem is as follows:

[0168] ;

[0169] in, The control sequence to be optimized includes torque control commands for both the engine and the electric motor. Preset control time domain; For a moment The corresponding comprehensive optimization control cost function integrates four core optimization sub-items. The function of each sub-item is as follows: 1. Operation quality tracking sub-item: By penalizing the estimated load torque value Compared with expected value To mitigate deviations and ensure operational quality stability, the weight matrix... 1. Used to adjust the optimization priority of this item; 2. Driving speed tracking sub-item: through penalized vehicle speed estimation value With target value To minimize deviations and ensure accurate vehicle speed tracking, meeting the requirements of agricultural operations for uniform vehicle speed, the weight matrix... Used to adjust the priority of this optimization; 3. Control smoothness sub-item: Controls the rate of change of input through penalty. To avoid drastic power fluctuations caused by sudden changes in engine and motor torque, and to ensure smooth operation, a weight matrix is ​​used. Used to adjust the optimization priority of this item; 4. Instantaneous fuel efficiency sub-item: penalized based on the instantaneous fuel consumption model. Calculated fuel consumption rate, reducing system energy consumption, improving instantaneous energy efficiency, weight matrix Used to adjust the priority of this optimization. For the preset prediction time domain; This is the current control cycle; To predict the offset; To predict the target time; , These are the estimated load torque and vehicle speed at the target time, respectively; To control the rate of change of input, all steps are defined in accordance with the previous steps.

[0170] S32. Using the feedforward power allocation benchmark as the initial value of the control sequence to be optimized, determine the initial solution for rolling optimization.

[0171] In this embodiment of the invention, the rolling optimization of model predictive control requires a reasonable initial solution as the starting point for the solution. If the initial solution deviates from the feasible region, it will lead to slow convergence speed of the optimization solution or even getting trapped in local optima, failing to meet the real-time requirements of agricultural machinery operation. The feedforward power distribution benchmark generated in step S103 has completed torque distribution based on the current operating mode and ensures that the engine and motor operate in their respective high-efficiency ranges. It has clear physical meaning and operating condition adaptability. Using it as the initial value of the control sequence to be optimized can effectively narrow the feasible region of the optimization solution and reduce the computational complexity. Its core function is to improve the convergence speed and stability of the MPC optimization solution and ensure that the optimization results can quickly meet the current operating condition requirements.

[0172] The feedforward dynamic distribution benchmark generated in step S103 Directly used as the control sequence to be optimized The initial solution is expressed as: ;in, This is the initial solution for rolling optimization. For a moment right The feedforward control reference at time (consistent with the definition of S103) has an initial solution that fully matches the load characteristics of the current working condition and the requirements for efficient operation of dual power sources. This can effectively avoid the blindness of optimization and improve optimization efficiency.

[0173] S33. Construct preset constraint conditions by adopting system dynamic constraints, control input constraints, and dynamic coordination strategy constraints.

[0174] In this embodiment of the invention, the solution of rolling optimization must be performed within a reasonable feasible region. Without constraints, the optimization results may exceed the system's physical limits or fail to meet operational requirements, leading to equipment damage or substandard operational quality. Therefore, it is necessary to construct three types of constraints based on the characteristics of the agricultural machinery system and operational needs: system dynamic constraints to conform to the actual motion laws of the agricultural machinery, ensuring that the optimization results conform to the vehicle's dynamic characteristics; control input constraints to limit the torque and speed limits of the engine and motor, avoiding exceeding the hardware's tolerance; and dynamic coordination strategy constraints to adapt to the current operating mode, clarifying the role allocation rules of the two power sources, and ensuring efficient collaboration between them. Its core function is to define the feasible region of optimization, ensuring the feasibility, safety, and adaptability of the optimization results, while simultaneously achieving dynamic coordination between the two power sources, balancing operational quality and system energy efficiency.

[0175] Furthermore, the dynamic coordination strategy constraints include motor-dominated mode, engine-dominated mode, and hybrid drive mode that match the current operating condition mode;

[0176] The constraints of the motor-dominated mode include that the torque borne by the motor is not less than a preset proportion of the total required torque, the engine operating range is limited to zero torque or a range that is not less than the preset lower limit of high-efficiency power generation torque and not higher than the preset upper limit of high-efficiency power generation torque, the remaining battery charge is not less than a preset low charge threshold, and the motor torque response rate is not less than the preset rate corresponding to the consistency requirements of agronomic operations.

[0177] The constraints of the engine-dominated mode include that the effective output power of the engine is not less than a preset multiple of the output power of the motor, the load borne by the engine is not less than the basic load that is positively correlated with the load intensity, the rate of change of engine torque does not exceed a preset safety threshold, and the motor only performs limited regenerative braking within the range that the battery charge is not less than a preset allowable threshold.

[0178] The constraints of the hybrid drive mode include that the proportion of motor power to total output power is not lower than a preset lower limit and not higher than a preset upper limit, and the engine working torque is not lower than the lower limit of the high-efficiency range corresponding to the optimal fuel consumption rate and not higher than the upper limit of the high-efficiency range.

[0179] In this embodiment of the invention, the specific details of the three types of preset constraints are as follows:

[0180] 1. System dynamics constraints: These are derived from the state estimator model in step S102, describing the evolution of the system state and ensuring that the optimization process closely matches the actual motion characteristics of the agricultural machinery. The expression is: ;in, , They are time points right , The predicted system state at time 1; Optimize decisions based on control inputs at the target time; Preview data (including waypoints and slope); The nonlinear functions are established based on the vehicle's longitudinal dynamics, rotational dynamics, and transmission relationships, and are consistent with the definitions in the previous steps.

[0181] 2. Control Input Constraints: Based on the hardware characteristics of the engine and motor, the physical limits of their torque output are defined to prevent equipment failure due to torque exceeding the range. The expression is: ;in, , These are the physical lower and upper limits of engine and motor torque, respectively, determined by the characteristics of the power source hardware and stored in the vehicle controller.

[0182] 3. Dynamic Coordination Strategy Constraints: These constraints, expressed in the form of inequality constraints, represent the coordination rules of the dual power sources under different operating conditions, adapting to the load characteristics and operational requirements of the current condition. Specifically, they are divided into three modes:

[0183] (1) Motor-dominated mode: For working conditions such as ramp transportation / transfer, the motor is forced to bear the main load, limiting the engine's operating range (zero torque or high-efficiency power generation range), while ensuring battery power and motor response rate, ensuring smooth operation and battery safety; the corresponding constraints are expressed as:

[0184] ;

[0185] in, For the current control cycle Next target time The optimized motor output torque; The motor torque distribution adjustment coefficient is used to force the motor to bear a preset proportion of the total required torque; For the target time Total torque demand; For the current control cycle Next target time The optimized engine output torque; This is the lower limit of the engine's high-efficiency power generation torque; This is the upper limit of the engine's efficient power generation torque. This means that the engine either outputs zero torque or operates within the range of high-efficiency torque generation. For the current control cycle Next target time Predicted remaining battery capacity; Set a low battery threshold to prevent over-discharge of the battery; For the target time The rate of change of the motor torque reflects the dynamic response performance of the motor; It is a dynamic torque adjustment coefficient used to match the consistency requirements of agronomic operations and to limit the minimum level of the motor torque response rate; For the target time The rate of change of total demand torque reflects the dynamic fluctuation characteristics of the load, which is consistent with the definition in the previous steps.

[0186] (2) Engine-dominated mode: For heavy load conditions such as rotary tillage and branch shredding, ensure that the engine bears the main load, limit the rate of change of engine torque, and limit the applicable conditions of regenerative braking of the motor to avoid inefficient engine operation and overcharging of the battery; the corresponding constraints are expressed as:

[0187] ;

[0188] in, This is the engine torque adjustment coefficient, used to adjust the effective output power of the engine; This is the motor power regulation coefficient, and... The ratio between engine output power and electric motor output power is determined accordingly. This represents the engine's basic torque, which is the minimum basic load the engine must withstand. This is the engine load distribution factor, which ensures that the load borne by the engine is positively correlated with the total required torque intensity. The engine torque change rate is set as a safe threshold to limit engine torque fluctuations and ensure smooth output. This is a battery power limiting function, with a value ranging from 0 to 1, which dynamically changes based on the remaining battery power. This is the maximum regenerative braking torque of the motor, limiting the motor to perform limited regenerative braking only when the battery power allows; the meanings of the other parameters are consistent with the corresponding parameters in the motor dominant mode constraints, and are all consistent with the definitions in the previous steps.

[0189] (3) Hybrid Drive Mode: For operating conditions such as weeding / light-load medium-high speed, the range of motor power ratio is limited to ensure that the engine operates within the optimal fuel consumption range, achieving efficient operation of the dual power sources. The corresponding constraints are expressed as follows:

[0190] ;

[0191] The first item specifies the upper and lower limits of the electric motor power ratio, defining a "gray area" for hybrid drive; the second item guides the engine's operating torque range to its fuel consumption rate. To ensure high efficiency at the system level, the optimal region is located nearby. The lower limit of the power allocation for hybrid drive is defined as the lower limit of the proportion of motor power. The upper limit of the power ratio of motors is set to define the upper limit of power allocation for hybrid drives; For the current control cycle Next target time The optimized motor output power; For the current control cycle Next target time The optimized engine output power; This is the lower limit of the torque for optimal fuel consumption of the engine, which limits the lower limit of the torque for efficient engine operation. The upper limit of torque for optimal fuel consumption rate of the engine limits the upper limit of torque for efficient engine operation. , This indicates that the engine needs to operate within the high-efficiency torque range corresponding to the optimal fuel consumption rate; the meanings of the remaining parameters are consistent with the corresponding parameters in the aforementioned mode constraints, and are all consistent with the definitions in the preceding steps.

[0192] S34. Under preset constraints, based on the multi-objective optimization function and the initial solution, rolling optimization is performed through the model predictive control algorithm to solve the finite-time domain optimal control problem within the preset prediction time domain, and to obtain the target control sequence within the preset control time domain.

[0193] In this embodiment of the invention, the core of model predictive control is "rolling optimization and real-time updating," and its core logic is in each control cycle. Starting with the initial solution determined in step S32, and within the constraints established in step S33, the optimal control sequence within the preset control time domain is solved by minimizing the multi-objective optimization function established in step S31. Since the load and road conditions are dynamically changing during agricultural machinery operation, the optimization result of a single cycle cannot adapt to subsequent working conditions. Therefore, this optimization process needs to be repeated in each control cycle to continuously update the optimal control sequence. Its core function is to achieve real-time and adaptive optimization decisions, ensuring that an optimal control scheme that takes into account multiple objectives can always be obtained in a dynamically changing operating environment, providing a basis for the torque command output of the current control cycle.

[0194] The rolling optimization process of the model predictive control algorithm is as follows: First, the initial solution... Substitute into the multi-objective optimization function Under the combined constraints of system dynamics, control input, and dynamic coordination strategy, the comprehensive optimization control cost is minimized using numerical solution algorithms (such as the interior-point method). Subsequently, the preset control time domain is obtained. Optimal control sequence within ;in For a moment right The optimal control input at any given time is the optimal torque command for both the engine and the electric motor. This optimization process occurs in each control cycle. Execute in real time to ensure that the optimization results can dynamically adapt to changes in the working environment and load.

[0195] S35. Take the first control element of the target control sequence as the target torque command for the engine and motor in the hybrid agricultural machinery within the current control cycle.

[0196] In this embodiment of the invention, the target control sequence obtained by model predictive control is the optimal control scheme for multiple future cycles within a preset control time domain. However, the torque control of agricultural machinery needs to be executed in real time; each control cycle only needs to execute the control command for the current moment, without needing to execute commands for future cycles in advance. Therefore, the first control element (i.e., the optimal control input corresponding to the current control cycle k) is extracted from the optimized target control sequence as the target torque command for the engine and motor in the current cycle. Its core function is to realize the MPC core logic of "rolling optimization and real-time execution," ensuring that the torque command for the current control cycle is globally optimal, while reserving space for optimization in the next control cycle, thus ensuring the continuity and real-time nature of the entire control process. The target control sequence obtained from step S34... Extract the first control element from the middle. The target torque command for the current control cycle k is expressed as: ;in, ,in, The target torque command for the engine in the current control cycle k; The target torque command for the motor in the current control cycle k is directly sent to the subsequent dual-loop PID controller to perform precise torque tracking control: the outer loop is the torque / speed loop, used to eliminate steady-state errors and achieve precise tracking of the target torque / speed; the inner loop is the current loop, used to achieve rapid dynamic tracking of torque. Finally, the inner loop controller outputs the physical signal that directly drives the actuator, controlling the engine fuel injection quantity and the motor drive current.

[0197] In embodiments of the present invention, such as Figure 3 As shown, this hybrid intelligent agricultural machine obtains and measures vehicle speed through a real-time data acquisition module and relies on an IMU sensor. Torque sensor acquires measured load torque LiDAR obtains the target road slope With road surface roughness The system combines operating condition information to complete multi-source data input. Then, the preview data, operating condition data, and noisy operating status data are input into an adaptive state observer composed of an operating condition discrimination module and an extended Kalman filter. Through preview feedforward compensation and noise filtering, it outputs a high-precision state estimate for the current control cycle. (Including actual driving speed, working load torque, and overall driving resistance, etc.); Then, based on the state estimate and target data, through dynamic coordination strategy, target model and feedforward module, the total torque required to maintain the target working quality and target driving speed is calculated, and a feedforward power distribution benchmark is generated according to the working condition adaptation rules. ; after that As the initial solution, under the constraints of system dynamics, control input, and dynamic coordination strategy (including three modes: motor-dominated, engine-dominated, and hybrid drive) to match the current operating conditions, a multi-objective optimization function is constructed through the MPC controller, focusing on operational quality stability, vehicle speed tracking accuracy, power switching smoothness, and instantaneous fuel efficiency. The system solves the finite-time optimal control problem online in a rolling manner to obtain the target control sequence within the preset control time domain. Finally, the first element of the target control sequence is taken as the target torque command for the engine and motor in the current cycle. This command is then converted into an execution signal by a dual-closed-loop PID controller to drive the engine and motor. After execution, the new state is collected by the sensor and fed back to the state estimation module, starting a new control cycle. This enables advanced, adaptive, and coordinated control of driving and operating power in complex farmland scenarios, optimizing the uniformity of operation, fuel economy, and smoothness of operation.

[0198] Example 2

[0199] This embodiment uses three typical working conditions—rotary tillage on flat land, slope operation, and high-speed transfer—as examples to illustrate the power compensation and coordinated control process of hybrid intelligent agricultural machinery. The initial state of the agricultural machinery is rotary tillage on flat land, and the central controller executes the following steps in a loop:

[0200] Step 1: Real-time sensing and input of multi-source information

[0201] The IMU and torque sensor collect the following information at a frequency of 100Hz: forward aiming data and operational status data. The LiDAR acquires the road gradient sequence and pavement roughness grade from the forward aiming data within 50m ahead; operational status data includes vehicle speed. Engine speed Motor speed Output shaft load torque .

[0202] Step 2: Real-time estimation of key states based on model and fusion

[0203] Using an adaptive state observer with a pre-aiming feedforward input channel, perform the following operations:

[0204] (1) State advance prediction: The road slope sequence in the advance data is used as a known disturbance input. The state advance prediction with advance feedforward is completed through the system state prediction formula to generate the system state prediction value, so that the prediction has feedforward characteristics.

[0205] (2) Quantification of prediction uncertainty: The prediction uncertainty is quantified by the formula for calculating the prediction state error covariance matrix, and the prediction state error covariance matrix with pre-aiming disturbance compensation is generated at the current time.

[0206] (3) Adaptive weight calculation: Based on the predicted state error covariance matrix, adaptive weight calculation is completed through the Kalman gain calculation formula to generate a Kalman gain that matches the current working condition;

[0207] (4) Deviation Correction and Noise Filtering: Based on Kalman gain, system state prediction, and sensor measurement vectors, deviation correction and iterative noise filtering are performed using the state update formula. The output state estimate for the current control cycle is then calculated. This includes vehicle speed estimates, load torque estimates, engine speed estimates, and motor speed estimates.

[0208] Step 3: Generate a feedforward reference based on preview and estimation

[0209] Based on a feedforward control model (built by coupling the longitudinal dynamics model of agricultural machinery and the load model of the implements), in the preset prediction time domain The total demand torque is calculated using the total demand torque calculation formula. Based on the torque distribution rules adapted to the current operating mode, the total demand torque is decomposed into the engine feedforward torque reference and the motor feedforward torque reference. The single-moment feedforward reference is obtained through the single-moment feedforward reference vector formula, and finally a complete feedforward control sequence is formed, namely the feedforward power distribution reference, which serves as the initial solution for subsequent optimization.

[0210] Step 4: Dynamic optimization decision under multi-objective constraints

[0211] In each control cycle k, construct and solve the Model Predictive Control (MPC) optimization problem:

[0212] (1) Optimization objective: The optimization objective is determined by a multi-objective optimization function. The objective function includes four sub-items: operation quality tracking, vehicle speed tracking, control smoothness, and instantaneous fuel efficiency. The core objective is to minimize the overall optimization control cost.

[0213] (2) Preset constraints: These include system dynamic constraints, control input constraints, and dynamic coordination strategy constraints that match the current operating conditions, wherein:

[0214] In rotary tillage operations, the engine-dominant mode constraint is activated to force the engine to bear the main load, ensuring stable operation quality.

[0215] In climbing conditions, the hybrid drive mode is switched to a constraint, limiting the range of motor power ratio and guiding the engine to operate in the optimal fuel consumption range to achieve a smooth power increase.

[0216] In the case of switching to the motor-driven mode, the motor is forced to bear the main load, and the engine is limited to operating only in the high-efficiency power generation range or at zero torque, in order to pursue the best fuel economy.

[0217] (3) Solution Output: The finite-time optimal control problem is solved online using a numerical optimization solver to obtain the target control sequence within the preset control time domain. The first control element of the target control sequence is taken, and the optimal torque command for the current control cycle is output. That is, the engine target torque command and the electric motor target torque command.

[0218] Step 5: Dual-loop precise tracking execution

[0219] The optimized engine target torque command and motor target torque command are sent to independent dual-closed-loop PID controllers for execution: the outer loop is the torque / speed loop, which calculates the current command based on the deviation between the command and the measured value to eliminate steady-state error; the inner loop is the current loop, which controls the motor driver or engine ECU to achieve rapid dynamic tracking of torque.

[0220] Step 6: Closed-loop feedback and adaptive iteration

[0221] After the actuator moves, the new vehicle and operation status is collected by the sensors and used as real-time status information for the next control cycle. This information is then fed back to the adaptive state observer with a preview feedforward input channel in step 2, initiating a new cycle of state estimation-feedforward prediction-optimization decision-execution feedback, thus forming a closed-loop adaptive control.

[0222] Furthermore, the dynamic adjustment mechanism under the three typical working conditions of rotary tillage, uphill climbing, and transportation is as follows:

[0223] (1) Steady-state rotary tillage operation on flat ground: The end load torque is large and stable. The core control objective is to maintain a uniform working depth and maintain vehicle speed. The adaptive state observer with a preview feedforward input channel makes predictions based on the large mass and high rotational inertia parameters corresponding to the rotary tillage mode. Since the preview data shows that the road ahead is flat, the road slope sequence in the preview data is... With the slope term in the disturbance input zero, the observer primarily filters sensor noise, outputting high-precision load torque and vehicle speed estimates. At this point, the comprehensive driving resistance estimate mainly includes rolling resistance and air resistance. The feedforward module, based on the feedforward control model, calculates the total torque required to maintain the current vehicle speed and target tillage depth using the total demand torque calculation formula. According to the characteristics of heavy-load steady-state operation, an engine-dominated mode is adopted, generating a feedforward power distribution benchmark through the engine and motor feedforward torque benchmark allocation formula. , ,in, This means the engine handles approximately 80% of the total torque demand, while the electric motor primarily compensates for minute, instantaneous fluctuations. Cost function In the context, the weight of the work quality item (load torque error) Set to maximum, vehicle speed tracking item Secondly, smoothness item and fuel economy The weight is low. Because... Continuously above the overload threshold Dynamic coordination strategy for activating engine-dominant mode in MPC:

[0224] ;

[0225] This constraint acts as a hard boundary, ensuring that any optimized solution places the engine under the primary load. Using a feedforward reference as initial values, the MPC iteratively solves for the optimal sequence over the next few seconds, satisfying the above constraints and system physical limits, and outputs the final torque command. Since the operating conditions are stable, the optimized command is very close to the feedforward reference, requiring only fine-tuning to optimize overall efficiency.

[0226] (2) Climbing operation condition: When the pre-aiming system detects that a continuous uphill section begins 20 meters ahead, the slope angle in the road slope sequence in the pre-aiming data increases from 0 to the value corresponding to a 15° slope. The adaptive state observer with the pre-aiming feedforward input channel is the first to sense the change and updates the road slope sequence in the pre-aiming data as a known disturbance. Input the state equation, and the observer's prediction step begins calculating the slope resistance. The impact on future states, even if the current vehicle speed has not yet decreased, includes the estimated output vehicle speed. and comprehensive driving resistance estimate The system has begun to reflect the impending increase in resistance, achieving proactive state awareness. The operating condition identification module determines that the current scenario is a medium-to-high load, switches the torque distribution rule to hybrid drive / climbing mode, and calculates the total required torque. At that time, the slope resistance term becomes dominant, and the allocation rule is adjusted to... , ;in, This means the engine handles approximately 60% of the total torque demand, while the feedforward power distribution benchmark includes a torque ramp-up sequence to increase power before the vehicle contacts the incline. In the multi-objective optimization function, the weight of the work quality tracking term decreases due to the reduced load, the weight of the vehicle speed tracking term is significantly increased to prevent rollback or sudden speed drops, and the weight of the control smoothness term is increased to optimize the control sequence for incline start-up. The hybrid drive mode constraint is triggered in the MPC, the original engine-dominated constraint is relaxed or replaced, and the new constraint set ensures sufficient resultant force and reasonable thermal load distribution. The MPC uses the feedforward incline compensation sequence as initial values ​​and re-optimizes under the new objective weights and constraint set. The output command causes the engine and motor to collaboratively increase torque, and due to the constraints, the distribution ratio remains within a reasonable range, achieving a smooth and powerful climb.

[0227] (3) Transfer Operation Condition: When the agricultural machinery finishes its operation and enters the transfer mode, the forward-looking data shows that there is a long, straight, hardened road ahead, the implements are completely suspended, and the load is very light. The adaptive state observer with the forward-looking feedforward input channel updates the disturbance input. With the gradient term set to zero, the system model weights shifted from overcoming gradient resistance to balancing rolling and air resistance. This resulted in a significant decrease in the estimated overall driving resistance, identifying the operating condition as light-load, high-speed transport, triggering the motor-dominated mode. The feedforward power distribution benchmark rule changed to the engine feedforward torque benchmark. Or a very small power generation demand: The motor feedforward torque reference is approximately equal to the total required torque, which is mainly used to overcome air resistance at high speeds. In the multi-objective optimization function, the instantaneous fuel efficiency term is set to the highest weight, the vehicle speed tracking weight is maintained, and the work quality tracking term weight is reduced to the lowest weight. Under the goal of minimizing fuel consumption, the MPC fully utilizes the motor's drive efficiency, outputting commands to shut down the engine or operate it in the most efficient power generation range, allowing the motor to independently drive the vehicle, thus achieving economical operation during field transitions.

[0228] This invention is a continuous adaptive process that uses preview data as a leading signal and operating condition identification as the decision-making basis, driving the state estimation model, feedforward rules, optimization objectives, and policy constraints layer by layer for collaborative reconstruction. This enables the control system not only to respond to existing state changes, but also to proactively, smoothly, and optimally prepare and execute control strategies for known future changes, thereby coping with complex farmland operation scenarios.

[0229] Example 3

[0230] Please see Figure 4 , Figure 4 This is a structural block diagram of a hybrid power compensation and cooperative control system for agricultural machinery provided in an embodiment of the present invention.

[0231] This invention provides a hybrid power compensation and cooperative control system for agricultural machinery, applicable to hybrid agricultural machinery. The system includes:

[0232] The multi-source data acquisition module 401 is used to acquire the pre-aiming data, working condition data and operating status data of the hybrid agricultural machinery.

[0233] The preview feedforward state estimation module 402 is used to perform feedforward compensation and noise filtering on preview data, operating condition data and running state data through an adaptive state observer with preview feedforward input channel, and generate the state estimation value of the current control cycle.

[0234] The working condition adaptation feedforward allocation module 403 is used to calculate the total required torque to maintain the target operation quality and target travel speed based on the state estimate and the aiming data, and to allocate the total required torque according to the current working condition mode to generate a feedforward power allocation benchmark.

[0235] The multi-objective rolling optimization module 404 is used to perform rolling optimization through a model predictive control algorithm under preset constraints, using the feedforward power distribution benchmark as the initial solution, to obtain the target torque commands of the engine and motor in the hybrid agricultural machinery.

[0236] In this embodiment of the invention, a hybrid power compensation and coordination control system for agricultural machinery is applied to hybrid agricultural machinery. The system includes a multi-source data acquisition module 401, a pre-aiming feedforward state estimation module 402, a working condition adaptation feedforward allocation module 403, and a multi-objective rolling optimization module 404, along with supporting actuators and a power supply system. Its core lies in constructing a closed-loop control architecture of pre-aiming feedforward – state estimation – dynamic coordination – multi-objective optimization: the multi-source data acquisition module 401 serves as the environmental perception entry point, acquiring pre-aiming data, working condition data, and operating status data through a sensor system; the pre-aiming feedforward state estimation module 402 serves as the core of information fusion and state estimation, using an adaptive state observer with a pre-aiming feedforward input channel to perform feedforward compensation and noise filtering on the multi-source data, accurately estimating key states such as vehicle speed and load torque in real time, and generating the state estimate value for the current control cycle; the working condition adaptation feedforward allocation module 403 serves as the core of feedforward predictive control, based on the state… The estimated value and the pre-aiming data are used to calculate the total torque required to maintain the target operation quality and target driving speed in advance based on the dynamics and load model, and to complete the power distribution according to the current working condition mode, generating a feedforward power distribution benchmark. The multi-objective rolling optimization module 404, as the core of multi-objective optimization decision, uses the feedforward power distribution benchmark as the initial solution. Under the constraints of the power system physical constraints and the preset dynamic coordination strategy of the engine and motor, it takes operation quality, vehicle speed tracking, power smoothness and energy efficiency as comprehensive objectives, and solves the finite-time domain optimal control problem online through the model predictive control algorithm, and outputs the optimal engine and motor target torque command in the current control cycle. Finally, the bottom-level dual closed-loop control module accurately converts the torque command into an execution signal through the PID controller, drives the engine and motor to work together, and the new state after execution is collected by the sensor and fed back to the pre-aiming feedforward state estimation module 402, forming a closed-loop control that continuously adapts to complex farmland operation scenarios. It achieves advanced, adaptive, and coordinated control of the driving and operating power of agricultural machinery, optimizing the uniformity of operation, fuel economy, smoothness of operation, and system response speed, and is especially suitable for complex farmland operation scenarios such as slopes and variable resistance.

[0237] The specific implementation method of the agricultural machinery hybrid power compensation and cooperative control method is basically the same as the specific implementation of the above-mentioned refrigeration machine performance prediction system, and will not be repeated here.

[0238] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for compensation and coordinated control of hybrid power in agricultural machinery, characterized in that, The method, applied to hybrid agricultural machinery, includes: Acquire the pre-aiming data, working condition data, and operating status data corresponding to the hybrid agricultural machinery; An adaptive state observer with a pre-aiming feedforward input channel performs feedforward compensation and noise filtering on the pre-aiming data, the operating condition data, and the running state data to generate a state estimate for the current control cycle. Based on the state estimate and the pre-aiming data, the total required torque to maintain the target operation quality and target travel speed is calculated, and the total required torque is distributed according to the current operating mode to generate a feedforward power distribution benchmark. Using the feedforward power distribution benchmark as the initial solution, rolling optimization is performed through a model predictive control algorithm under preset constraints to obtain the target torque commands for the engine and motor in the hybrid agricultural machinery.

2. The agricultural machinery hybrid power compensation and cooperative control method according to claim 1, characterized in that, The step of generating a state estimate for the current control cycle by performing feedforward compensation and noise filtering on the pre-aiming data, the operating condition data, and the running state data using an adaptive state observer with a pre-aiming feedforward input channel includes: Based on the longitudinal dynamics, rotational dynamics, and transmission relationship of agricultural machinery, a nonlinear dynamic function matching the hybrid agricultural machinery is constructed, and an adaptive state observer with a pre-aiming feedforward input channel is built by combining the extended Kalman filter algorithm framework. The pre-aiming data, the operating condition data, and the running status data are standardized to generate system status data and sensor measurement vectors that are adapted to the adaptive state observer. The system state data, the future road condition disturbances in the forecast data, the control command of the previous cycle in the operating condition data, and the state estimate of the previous cycle are input into the forecast feedforward input channel of the adaptive state observer to perform state advance prediction and generate system state prediction values. The adaptive state observer performs prediction uncertainty quantification and adaptive weight calculation based on the system state prediction value, the Jacobian matrix of the nonlinear dynamic function, the error covariance matrix of the previous period, the preset process noise covariance matrix, and the preset observation noise covariance matrix, and generates a Kalman gain that matches the current operating condition. Based on the Kalman gain, the system state prediction value, and the sensor measurement vector, deviation correction and iterative noise filtering are performed to generate the state estimate value for the current control cycle.

3. The agricultural machinery hybrid power compensation and cooperative control method according to claim 2, characterized in that, The step of generating a Kalman gain that matches the current operating condition by quantifying prediction uncertainty and calculating adaptive weights using the adaptive state observer, based on the predicted system state value, the Jacobian matrix of the nonlinear dynamic function, the error covariance matrix of the previous period, the preset process noise covariance matrix, and the preset observation noise covariance matrix, includes: Extract the Jacobian matrix of the nonlinear dynamic function at the system state prediction value with pre-aiming feedforward compensation; Using the Jacobian matrix, the error covariance matrix of the previous period, and the preset process noise covariance matrix, the prediction state error covariance matrix with pre-aiming disturbance compensation at the current time is calculated. Based on the predicted state error covariance matrix, the observation matrix, and the preset observation noise covariance matrix, an adaptive weighting coefficient is calculated to balance the confidence of the aiming prediction and the confidence of the sensor measurement, thereby generating a Kalman gain that matches the current operating conditions.

4. The agricultural machinery hybrid power compensation and cooperative control method according to claim 2, characterized in that, The step of performing deviation correction and iterative noise filtering based on the Kalman gain, system state prediction value, and sensor measurement vector to generate the state estimate value for the current control cycle includes: Using the sensor measurement vector and the system state prediction value, the measurement residual between the sensor measured value and the system state prediction value is calculated to obtain the measurement residual deviation at the current moment; The measurement residual bias is weighted using the Kalman gain to obtain the state correction increment at the current moment; The system state prediction value is corrected for deviation using the state correction increment to generate intermediate state correction results. The intermediate state correction result is iteratively filtered to suppress the observation noise introduced by the sensor measurement vector and generate the state estimate value within the current control cycle.

5. The agricultural machinery hybrid power compensation and cooperative control method according to claim 1, characterized in that, The step of calculating the total required torque to maintain the target operation quality and target travel speed based on the state estimate and the pre-aiming data, and distributing the total required torque according to the current operating mode to generate a feedforward power distribution benchmark, includes: With the goal of maintaining the target operation quality and target travel speed, a feedforward control model is determined based on the longitudinal dynamics model of agricultural machinery and the load model of the operating implements; The feedforward control model performs forward calculations on the state estimate, the future path information and road slope sequence in the forward data within a preset prediction time domain to generate the total demand torque corresponding to each moment in the prediction time domain. Based on the principle that both the engine and motor in the hybrid agricultural machinery operate within their respective high-efficiency working ranges, a torque distribution rule adapted to the current working condition is determined. According to the torque distribution rule, the total required torque is distributed to obtain the engine feedforward torque reference and the motor feedforward torque reference, and a feedforward power distribution reference is generated.

6. The agricultural machinery hybrid power compensation and cooperative control method according to claim 5, characterized in that, The steps, which are based on the principle that both the engine and motor in the hybrid agricultural machinery operate within their respective high-efficiency working ranges, and are adapted to the torque distribution rules of the current operating mode, include: Based on the universal characteristic curve corresponding to the universal characteristic data of the engine in the operating condition data, the optimal fuel consumption rate range is calibrated, and the threshold of the engine's high-efficiency operating range is generated. Based on the rated operating parameters of the motor in the operating condition data, the rated high-efficiency operating range is calibrated, and the threshold of the high-efficiency operating range of the motor is generated. Using the engine's high-efficiency operating range threshold and the motor's high-efficiency operating range threshold, combined with the hybrid power system characteristic data, agronomic operation demand data, and agricultural machinery industry engineering practice data in the operating condition data, a core principle for torque distribution is formulated, generating a torque distribution principle with the engine and motor in the hybrid agricultural machinery operating in their respective high-efficiency operating ranges as the core. Perform operating condition feature identification on the current operating condition data to generate load characteristic parameters and industry standard requirement parameters corresponding to the current operating condition. Based on the torque distribution principle, the load characteristic parameters, the industry standard requirement parameters, the engine high-efficiency operating range threshold, and the motor high-efficiency operating range threshold, a torque distribution rule adapted to the current operating condition is constructed.

7. The agricultural machinery hybrid power compensation and cooperative control method according to claim 6, characterized in that, The step of constructing a torque distribution rule adapted to the current operating condition based on the torque distribution principle, the load characteristic parameters, the industry standard requirement parameters, the engine high-efficiency operating range threshold, and the motor high-efficiency operating range threshold includes: Based on the load characteristic parameters and the industry standard requirement parameters, an initial candidate rule adapted to the current operating mode is constructed. Based on the engine's high-efficiency operating range threshold and the motor's high-efficiency operating range threshold, the initial candidate rules are checked for adaptability, rules that do not meet the preset requirements for efficient operation of dual power sources are eliminated, and target candidate rules are generated. Based on the torque distribution principle, the target candidate rules are fine-tuned and optimized to generate torque distribution rules that are adapted to the current operating mode.

8. The agricultural machinery hybrid power compensation and cooperative control method according to claim 1, characterized in that, The step of obtaining the target torque command for the engine and motor in the hybrid agricultural machinery by performing rolling optimization through a model predictive control algorithm under preset constraints, using the feedforward power distribution benchmark as the initial solution, includes: A multi-objective optimization function is constructed with the goal of minimizing the comprehensive optimization control cost within the preset prediction time domain; Using the feedforward power allocation benchmark as the initial value of the control sequence to be optimized, the initial solution for rolling optimization is determined; Pre-defined constraint conditions are constructed by employing system dynamics constraints, control input constraints, and dynamic coordination strategy constraints. Under the preset constraints, based on the multi-objective optimization function and the initial solution, rolling optimization is performed by the model predictive control algorithm to solve the finite-time domain optimal control problem within the preset prediction time domain, thereby obtaining the target control sequence within the preset control time domain. The first control element of the target control sequence is taken as the target torque command for the engine and motor of the hybrid agricultural machinery in the current control cycle.

9. The agricultural machinery hybrid power compensation and cooperative control method according to claim 8, characterized in that, The dynamic coordination strategy constraints include motor-dominated mode, engine-dominated mode, and hybrid drive mode that match the current operating mode. The constraints of the motor-dominated mode include that the torque borne by the motor is not less than a preset proportion of the total required torque, the engine working range is limited to a range of zero torque or not less than a preset lower limit of high-efficiency power generation torque and not more than a preset upper limit of high-efficiency power generation torque, the remaining battery power is not less than a preset low power threshold, and the motor torque response rate is not less than a preset rate corresponding to the consistency requirements of agronomic operations. The constraints of the engine-dominated mode include: the effective output power of the engine is not less than a preset multiple of the output power of the motor; the load borne by the engine is not less than the basic load that is positively correlated with the load intensity; the rate of change of engine torque does not exceed a preset safety threshold; and the motor only performs limited regenerative braking within the range where the battery charge is not less than a preset allowable threshold. The constraints of the hybrid drive mode include that the proportion of motor power to total output power is not lower than a preset lower limit and not higher than a preset upper limit, and the engine working torque is not lower than the lower limit of the high-efficiency range corresponding to the optimal fuel consumption rate and not higher than the upper limit of the high-efficiency range.

10. A hybrid power compensation and cooperative control system for agricultural machinery, characterized in that, The system, applied to hybrid agricultural machinery, includes: The multi-source data acquisition module is used to acquire the pre-aiming data, working condition data and operating status data corresponding to the hybrid agricultural machinery; The preview feedforward state estimation module is used to perform feedforward compensation and noise filtering on the preview data, the operating condition data and the running state data through an adaptive state observer with a preview feedforward input channel, and generate the state estimation value of the current control cycle. The working condition adaptation feedforward allocation module is used to calculate the total required torque to maintain the target operation quality and target travel speed based on the state estimate and the pre-aiming data, and to allocate the total required torque according to the current working condition mode to generate a feedforward power allocation benchmark. The multi-objective rolling optimization module is used to perform rolling optimization using the feedforward power distribution benchmark as the initial solution and under preset constraints through a model predictive control algorithm to obtain the target torque commands of the engine and motor in the hybrid agricultural machinery.

Images