Electro-hydraulic power source variable-speed variable-displacement control method capable of minimizing global energy consumption

Abstract

The invention discloses a variable-speed variable-displacement control method for an electro-hydraulic power source with the global energy consumption minimized. The variable-speed variable-displacement control method comprises the following steps that firstly, operation data of a hydraulic system under multiple working conditions are obtained; 2, constructing a system input power prediction model, and learning a nonlinear mapping relation between operation data and input power; 3, discretizing the continuously changing working conditions, and dividing the working conditions into a plurality of discrete time periods; 4, in each discrete time period, generating a candidate state set of the motor rotating speed and the pump displacement, and predicting the input power of each candidate state by using the power prediction model; 5, based on the dynamic planning principle, calculating the minimum accumulative operation cost of each time period, and generating a global optimal rotating speed and displacement sequence meeting dynamic constraint conditions; and 6, starting from the initial state, positively reconstructing the globally optimal motor rotating speed and pump displacement sequence to serve as reference input of a control system, and achieving global energy consumption minimization operation of the hydraulic system.

Description

Technical Field

[0001] The present invention belongs to the technical field of hydraulic transmission, and specifically relates to a variable speed and variable displacement control method for an electro-hydraulic power source with minimized global energy consumption. Background Art

[0002] With the increasing demand for energy conservation and efficient operation, hydraulic systems driven by variable-speed motors and variable-displacement pumps have been widely used in engineering machinery, industrial equipment, and new energy fields. By coordinating the motor speed and pump displacement, the system output flow can be flexibly controlled, and the motor and pump can be operated in the high-efficiency zone, thus having the potential for low energy consumption, fast response, and high control accuracy. In existing research, common energy-saving control methods mainly include: fixed control strategies based on empirical laws, optimal operating point selection based on efficiency maps, and energy distribution under given load conditions using genetic algorithms or other optimization algorithms. These methods can achieve certain energy-saving effects under static or quasi-static conditions, providing useful ideas for the optimization of hydraulic drive systems.

[0003] However, existing technologies are generally based on the assumption of discrete working conditions, that is, it is assumed that the system load remains unchanged for a long time, and then optimization is performed under this working condition; when the working conditions change, the new working condition is reset and the optimization is repeated. This method is suitable for scenarios where the working conditions change slowly or are segmented and constant. However, in actual engineering applications, the load often presents continuous dynamic changes, and optimization based only on discrete working conditions is difficult to ensure optimal energy consumption during the overall operation process. In addition, some methods ignore the dynamic response constraints of the motor and pump during the optimization process, assuming that the speed and displacement can jump arbitrarily, making the optimization solution difficult to achieve in actual operation; and if the complete dynamic equation is directly introduced, the computational complexity will be too high, making it difficult to meet the needs of real-time optimization. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a variable speed and variable displacement control method for an electro-hydraulic power source with minimized global energy consumption, which can take into account both dynamic response characteristics and computational efficiency, so as to achieve optimal energy consumption control of a variable speed motor and variable displacement pump combined drive system under continuously changing working conditions.

[0005] In order to achieve the above object, the present invention provides the following technical solutions: A method for controlling a variable speed and displacement of an electro-hydraulic power source with global energy consumption minimization includes the following steps: Step 1: Obtain the operating data of the hydraulic system under multiple working conditions; Step 2: Build a system input power prediction model based on the extreme gradient boosting algorithm to learn the nonlinear mapping relationship between the operating data and the input power; Step 3: Discretize the continuously changing working conditions and divide them into multiple discrete time periods; Step 4: In each discrete time period, a set of candidate states of motor speed and pump displacement is generated, and the input power of each candidate state is predicted using the power prediction model; the generation of the candidate state set is limited to a dynamic change range determined by the maximum rate of change of motor speed and the maximum rate of change of pump displacement; Step 5: Based on the principle of dynamic programming, the minimum cumulative operating cost and the corresponding optimal control variables of each time period are calculated through reverse recursion to generate the global optimal speed sequence and displacement sequence that meet the dynamic constraints; Step 6: Starting from the initial state, forward reconstruct the globally optimal motor speed sequence and pump displacement sequence as the reference input of the control system to achieve global energy minimization operation of the hydraulic system.

[0006] Furthermore, in step 1, the method steps for obtaining the operation data include: 11) Get the motor speed , outlet pressure , export traffic ,temperature and system input power , and based on the physical relationship of the pump port Derived pump displacement; 12) Preprocess the running data, including removing outliers and normalizing them, and use a random partitioning strategy to divide them into training and test sets.

[0007] Furthermore, in step 2, the step of constructing the input power prediction model includes: 21) Build an input power prediction model using an extreme gradient boosting algorithm regression architecture, including: accelerating computation through a data matrix structure; utilizing a histogram tree construction method for efficient memory management; determining the optimal combination through hyperparameter grid search; and enabling multi-core parallel training to improve training efficiency. 22) Use the coefficient of determination And root mean square error RMSE evaluate model performance, including: determination coefficient is greater than a preset first threshold, and the root mean square error RMSE is less than a preset second threshold.

[0008] Furthermore, in step 3, the time step of the discretization process is The value range is 0.2-0.5 seconds to adapt to the dynamic response characteristics of pump displacement and motor speed and reduce sensitivity to noise.

[0009] Furthermore, in step 4, the step of constructing the candidate state set includes: According to the dynamic adjustment capabilities of the motor and pump, the dynamic change range of speed and displacement is set to limit the state jump between adjacent time periods; In step 4, the method for constructing the candidate state set includes: setting the dynamic change interval of the speed and displacement according to the dynamic adjustment capabilities of the motor and pump, and limiting the state jump between adjacent time periods; the dynamic change interval is expressed as: in: For the The state at a moment; and Respectively Motor speed and pump displacement at each moment; For the The state at a moment; and Respectively Motor speed and pump displacement at each moment; and Respectively represent The maximum speed change and displacement change allowed in a discrete time step; is the discrete grid set of motor speed; is the minimum motor speed allowed / controllable; is the maximum motor speed allowed / controllable; is the discrete grid set of pump displacement; is the minimum allowable / controllable pump displacement; is the maximum allowable / controllable pump displacement; and are the speed discrete grid step and the displacement discrete grid step respectively.

[0010] Furthermore, in step 5, the method steps for calculating the minimum cumulative operating cost and the corresponding optimal control amount in each time period by reverse recursion are as follows: 51) Order , is the number of discrete points, For the The number of operating points in a discrete time period, and ; 52) Initialize the candidate state set for the first operating point , from the candidate state set Select Status , the input power prediction model is used to predict the state Input power under ; 53) Calculation status Next The operating cost of a discrete time period , operating costs It is obtained by accumulating the energy consumption of two adjacent operating points in time steps, and for the Any two adjacent discrete time periods Operating point and Energy consumption of the operating point time step, , and its calculation steps are: 531) Status-based Generate the Candidate state set of operating point , from the candidate state set Select Status , the input power prediction model is used to predict the state Input power under ; Among them: if , then the state Candidate state set The currently selected state ;like , then the state Candidate state set Medium energy consumption Minimal state ; 532) Based on the adjacent Operating point and The input power of the operating point is used to determine the energy consumption of the time step of the two adjacent operating points by multiplying the average power by the time step. ; 533) Traverse the candidate state set , get in state Energy consumption Minimal state And the corresponding minimum energy consumption ; 54) Traverse the candidate state set , get the The operating cost is set in a discrete time period The smallest state combination of each operating point, and mapping the state combination to the corresponding control variable sequence; 55) Judgment Is it greater than or equal to 0? If so: , looping through step 52); if so, the iteration ends, and the optimal control variable sequence corresponding to the cumulative minimum operating cost is obtained, and the optimal control variable sequence is used as the global optimization control trajectory of the hydraulic system.

[0011] Furthermore, the control variable sequence includes a motor speed sequence and a pump displacement sequence.

[0012] Furthermore, the adjacent Operating point and The energy consumption at the operating point is: in: For the adjacent Operating point and The time step between load points.

[0013] Further, in step 53), The cumulative minimum operating cost at the operating point is expressed by the cumulative energy consumption cost function: in: Indicates the discrete time periods in the The working point is always in the state The minimum cumulative operating cost when For the discrete time periods in the Optimal cumulative energy consumption at the operating point; Status Transfer to state The corresponding energy consumption cost; A pair of adjacent states.

[0014] Furthermore, in step six, the forward reconstruction step includes: Starting from the initial state, the optimal sequence of motor speed and pump displacement is gradually generated; The sequence is input into a power prediction model, the optimized total energy consumption is calculated, and the optimized total energy consumption is output as a control instruction.

[0015] The beneficial effects of the present invention are: The global energy consumption minimization electro-hydraulic power source variable speed variable displacement control method of the application is suitable for a variable speed variable displacement (SDVP) hydraulic system. The method of the application firstly uses an extreme gradient boosting algorithm (XGBoost) to construct an input power prediction model, learns the nonlinear mapping relationship between the hydraulic system operation data and the input power, and realizes high-precision prediction of energy consumption under different working conditions; then, based on the dynamic programming principle, the continuously changing working conditions are discretized into a multi-stage optimization problem, the total input energy consumption is minimized as the target, the power prediction model is combined for global optimization; finally, the optimal cost and optimal control amount of each stage state are obtained in the reverse recursion stage, and the globally optimal speed sequence and displacement sequence are generated in the forward reconstruction stage, which can effectively reduce the calculation complexity to meet the real-time optimization demand. In summary, the global energy consumption minimization electro-hydraulic power source variable speed variable displacement control method of the application can balance the dynamic response characteristics and calculation efficiency, so as to realize the energy optimal control of the variable speed motor and variable displacement pump combined driving system under continuously changing working conditions.

[0016] The application also has the following advantages: (1) By introducing the dynamic constraint interval of motor speed and pump displacement in the dynamic programming search, unrealistic jumps of state variables are avoided, the solving result is closer to the physical response characteristics of the actual system, and the realizability of the optimization strategy is improved; (2) The simplified method based on candidate interval restriction is proposed, which avoids directly solving complex differential equations, significantly reduces the calculation amount, improves the optimization solving efficiency while ensuring the rationality of dynamic constraints; (3) In the energy consumption calculation, the step-by-step energy consumption integration method under discrete time steps is adopted, which can more accurately reflect the instantaneous power consumption of the motor and pump under different working conditions, so as to obtain more accurate energy consumption evaluation results; (4) Through the reverse iterative dynamic programming strategy, the joint optimization control of motor speed and pump displacement is realized under the premise of meeting the flow demand and dynamic constraints, and the overall energy consumption of the system is effectively reduced. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to make the purpose, technical scheme and beneficial effects of the application clearer, the application provides the following drawings for illustration: Figure 1 The flowchart of the global energy consumption minimization electro-hydraulic power source variable speed variable displacement control method of the application. DETAILED DESCRIPTION

[0018] The application will be further described below in combination with the drawings and specific embodiments, so that those skilled in the art can better understand the application and implement it, but the embodiments are not limiting to the application.

[0019] As Figure 1 As shown, the variable speed and variable displacement control method of the electro-hydraulic power source with global energy consumption minimization in this embodiment includes the following steps: Step 1: Obtain the operating data of the hydraulic system under multiple working conditions.

[0020] Specifically, in this embodiment, the method and steps for obtaining operation data include the following contents.

[0021] 11) Obtain stable operating data of the hydraulic system under different outlet flow and pressure conditions, including motor speed , outlet pressure , export traffic ,temperature and system input power , and based on the physical relationship of the pump port The pump displacement is derived, eliminating the need for direct displacement data collection.

[0022] 12) Preprocess the operational data, including removing outliers and normalizing them. This involves removing data with significant input power deviations and normalizing them to eliminate dimensional differences. Finally, the preprocessed operational data is randomly partitioned into training and test sets.

[0023] Step 2: Build a system input power prediction model based on the extreme gradient boosting algorithm to learn the nonlinear mapping relationship between operating data and input power.

[0024] In this embodiment, the steps of constructing the input power prediction model include the following contents.

[0025] 21) Use the extreme gradient boosting algorithm regression architecture to build an input power prediction model, accelerate calculations through data matrix data structures; use the histogram tree construction method to achieve efficient memory management; determine the optimal combination through hyperparameter grid search; enable multi-core parallel training to improve training efficiency.

[0026] 22) Use the coefficient of determination And root mean square error RMSE evaluate model performance, including: determination coefficient The prediction model is larger than a preset first threshold, and the root mean square error (RMSE) is smaller than a preset second threshold. In this embodiment, the first threshold is set to 0.95, and the second threshold is determined based on the actual application scenario. The final model and normalizer are persistently stored, forming a deployable prediction pipeline.

[0027] Step 3: Discretize the continuously changing operating conditions into multiple discrete time periods. For each time period's state variables (including motor speed and pump displacement), a candidate state set is constructed within the system's constraints. The trained input power prediction model is then used to rapidly batch-predict the input power for each candidate state. Based on this, the minimum cumulative operating cost and the corresponding optimal control variable for each time period are calculated through reverse recursion, following the principles of dynamic programming. This results in a global optimal state-control strategy mapping that satisfies the dynamic constraints.

[0028] Specifically, the working flow and pressure Throughout the duration The internal discretization is into N time points, which can be expressed as: in: is the discretized time step.

[0029] In this embodiment, the operating flow and pressure of the hydraulic system during a continuous operating time interval are discretized to obtain multiple discrete time periods. Each discrete time period includes at least two operating points. When there are two operating points, the two operating points are the starting operating point and the ending operating point. When there are more than two operating points, the operating points include the starting operating point, the ending operating point, and at least one intermediate operating point between the starting and ending operating points. Each operating point contains information such as time, flow, and pressure.

[0030] In a variable speed and variable displacement system, the displacement adjustment mechanism of the variable pump is affected by mechanical inertia, friction resistance and control oil pressure dynamics. Its equivalent time constant is usually in the order of hundreds of milliseconds, which is significantly slower than the electromagnetic response of the motor. In combination with the control bandwidth of the system and the sampling theorem, in this embodiment, the time step of the discretization processing is selected The value range is 0.2-0.5 seconds, which can adapt to the dynamic response characteristics of pump displacement and motor speed, and reduce the sensitivity to noise while ensuring calculation margin.

[0031] Step 4: In each discrete time period, a set of candidate states for motor speed and pump displacement is generated, and the input power of each candidate state is predicted using the power prediction model. The generation of the candidate state set is limited to the dynamic change range determined by the maximum rate of change of motor speed and the maximum rate of change of pump displacement.

[0032] Specifically, the working condition set is expressed as: To calculate the energy consumption of a hydraulic system, the energy consumption is usually expressed as the integral of power over time within a time period: in: Indicates the instantaneous input power; For the Energy consumption in a time period.

[0033] Since the power curve is inevitably approximated between adjacent time points during the discretization process of dynamic programming, the trapezoidal integral approximation formula is used for calculation here: in: For the The input power at a certain time point; For the The input power at a given time point.

[0034] When searching for candidate motor speeds and pump displacements, the dynamic characteristics of the actuator must be considered. For a permanent magnet synchronous motor, its speed change is affected by the electromagnetic torque, load torque, and damping factors. The dynamic relationship can be expressed as: in: is the moment of inertia, is the angular velocity, is the electromagnetic torque, is the load torque, is the damping coefficient.

[0035] As can be seen from this equation, the acceleration and deceleration process of the motor is limited by the torque difference. Its speed change rate is not unlimited, but has an upper limit determined by the mechanical characteristics and control capabilities. Similarly, the displacement adjustment process of the variable pump is usually achieved by a proportional servo mechanism, and its dynamic characteristics can be simplified to a first-order inertia link: in: is the actual displacement, To control the command displacement, It can be seen that the change of pump displacement is also limited by the response speed of the actuator, and it is impossible for a large jump to occur at adjacent discrete moments.

[0036] In order to reflect the above physical constraints in the dynamic programming search and avoid the computational complexity brought by directly solving the differential equations, in this embodiment, the steps of constructing the candidate state set include: setting the dynamic change range of speed and displacement according to the dynamic adjustment capabilities of the motor and pump, and limiting the state jump in adjacent time periods; the dynamic change range is determined by the maximum speed change rate of the motor. and the maximum displacement change rate of the pump Sure: in: For the The state at a moment; and Respectively Motor speed and pump displacement at each moment; For the The state at a moment; and Respectively The motor speed and pump displacement at each moment.

[0037] Specifically, the maximum speed change rate of the motor and the maximum displacement change rate of the pump It is not a fixed constant, but rather a combination of the maximum allowable rate of change of the motor speed and the maximum allowable rate of change of the pump displacement. This approach significantly narrows the search space for candidate states while ensuring physical feasibility and, to a certain extent, reflects the dynamic response constraints of the motor and pump.

[0038] At each discrete time step According to the current pump outlet pressure Calculate the dynamic response performance of the motor and pump to obtain the maximum speed change rate that adapts to pressure and maximum displacement change rate , and accordingly set the dynamic change range of speed and displacement to limit the state jump between adjacent time periods: in: For the The state at a moment; and Respectively Motor speed and pump displacement at each moment; For the The state at a moment; and Respectively Motor speed and pump displacement at each moment; is the discrete time step; Determined by the torque margin-inertia constraint of the motor under pressure-related load, Determined by the pump's pressure-dependent servo / swashplate response; and: Maximum speed change rate Given by the following formula: in: is the equivalent moment of inertia; is the viscous damping coefficient; Indicates the maximum load torque of the pump; ; ; is the safety factor; For the The pump load torque at the state of the moment is expressed as: in: for mechanical efficiency; is the additional damping coefficient; Maximum displacement change rate Determined by the pressure-dependent first-order inertia swash plate model: in: Indicates the Pump outlet pressure at the moment (unit: MPa); Indicates the equivalent time constant of the first-order response of the variable displacement pump at this pressure (unit: s); and They represent the first-order derivative of displacement with respect to time (displacement change rate, unit: ) and instantaneous displacement (unit: ); Indicates the target displacement / swash plate target position given by the controller (unit: )); Indicates the reference time constant at zero pressure or reference pressure (unit: s); Indicates the sensitivity coefficient of time constant to pressure (unit: s / MPa); , is the safety factor; To adapt to the engineering control resolution, first calculate the decimal step size and , then do the nearest grid mapping (quantization): in: and Respectively represent The maximum speed change (step size after quantization) and maximum displacement change (step size after quantization) allowed in a discrete time step; and are the speed discrete grid step and the displacement discrete grid step respectively.

[0039] Therefore, the feasible interval of adjacent time periods is revised as follows: wherein: is a discrete grid set of motor speed; is the lowest allowable / controllable motor speed; is the highest allowable / controllable motor speed; is a discrete grid set of pump displacement; is the lowest allowable / controllable pump displacement; is the highest allowable / controllable pump displacement.

[0040] Step five: based on the dynamic programming principle, the minimum cumulative running cost and the corresponding optimal control variable of each time period are calculated by reverse recursion to generate the globally optimal speed sequence and displacement sequence that meet the dynamic constraint conditions.

[0041] For the time series flow and pressure data obtained by discretizing the working condition, first, the candidate grid of motor speed and pump displacement is generated at each discrete time point; the flow and pressure are input into the trained input power prediction model, and the instantaneous power under the candidate state is calculated combined with the current speed and displacement; the trapezoidal integration method is used to obtain the single-step energy consumption between adjacent time points. On this basis, the global optimal solution is obtained by using reverse dynamic programming: the cumulative energy consumption of each state variable is calculated by layer-by-layer backtracking, and the corresponding optimal control variable is updated synchronously. In order to ensure the dynamic realizability of the system, the state change range is limited in the iteration process: the candidate interval of motor speed is set as the symmetric neighborhood of the speed at the previous time, and its width is determined by the dynamic speed regulation ability of the motor; the candidate interval of pump displacement is set as the symmetric neighborhood of the displacement at the previous time, and its width is determined by the dynamic adjustment ability of the pump. If the candidate displacement exceeds the physical allowable interval, it is discarded; if the constraint condition is met, the single-step energy consumption is calculated based on the average power of the state and the next state, and is added to the subsequent cumulative energy consumption to update the minimum running cost of the current state. Finally, the path with the minimum total energy consumption is selected from all candidate states to obtain the optimal speed and displacement sequence.

[0042] In this embodiment, the method steps for calculating the minimum cumulative running cost and the corresponding optimal control variable of each time period by reverse recursion are: 51) Let , is the number of discrete points, is the number of working condition points in the th discrete time period, and .

[0043] 52) Initialize the candidate state set of the 1st working condition point , select state from the candidate state set , and use the input power prediction model to predict the state Input power under .

[0044] 53) Calculation status Next The operating cost of a discrete time period , operating costs It is obtained by accumulating the energy consumption of two adjacent operating points in time steps, and for the Any two adjacent discrete time periods Operating point and Energy consumption of the operating point time step, , and its calculation steps are as follows.

[0045] 531) Status-based Generate the Candidate state set of operating point , from the candidate state set Select Status , the input power prediction model is used to predict the state Input power under ; Among them: if , then the state Candidate state set The currently selected state ;like , then the state Candidate state set Medium energy consumption Minimal state In this embodiment, from the candidate state set Select Status When selecting the motor speed , and then use the physical relationship Derived pump displacement , and determine the calculated pump displacement Is it beyond the effective range of pump displacement? If so, the selected state is invalid and needs to be reselected from the candidate state set. Select the state; if not, the selected state is valid, and the state is predicted using the input power prediction model Input power under .

[0046] 532) Based on the adjacent Operating point and The input power of the operating point is used to determine the energy consumption of the time step of the two adjacent operating points by multiplying the average power by the time step. .

[0047] The adjacent Operating point and Energy consumption at operating point The calculation method is: in: For the adjacent Operating point and The time step between load points.

[0048] Based on the dynamic change interval restriction, the dynamic programming method is used to achieve the global optimal search. Specifically, starting from the task terminal moment, according to the load demand, reverse traverse to the initial moment, and gradually build the candidate state set. Candidate states at operating point , using dynamic intervals to filter the feasible state set at adjacent moments For each pair of adjacent states , through the single-step energy consumption calculation in step 532), the energy consumption cost corresponding to the state transition is obtained On this basis, The cumulative minimum operating cost at the operating point is expressed by the cumulative energy consumption cost function: in: Indicates the discrete time periods in the The working point is always in the state The minimum cumulative operating cost when For the discrete time periods in the Optimal cumulative energy consumption at the operating point; Status Transfer to state The corresponding energy consumption cost; A pair of adjacent states.

[0049] 533) Traverse the candidate state set , get in state Energy consumption Minimal state And the corresponding minimum energy consumption .

[0050] Through the step 53) of the step-by-step iteration, the candidate state set can be obtained. Select Status Under the conditions, The state of other operating points that minimize the energy consumption between two adjacent operating points in a discrete time period, but the state Can the first minimizing the operating cost in the first discrete time period further iteration is needed.

[0051] 54) traversing the candidate state set to obtain the state combination of each operating point that minimizes the operating cost in the first discrete time period, and mapping the state combination to the corresponding control variable sequence; specifically, the control variable sequence includes the motor speed sequence and the pump displacement sequence. In this way, by traversing the candidate state set , the minimum operating cost

[0052] under each state condition in the candidate state set can be obtained respectively. From these minimum operating costs , the minimum one is selected, that is, the state corresponding to the candidate state set and the state that minimizes the energy consumption of other operating points , so as to obtain the state combination of each operating point that minimizes the operating cost in the first discrete time period. 55) judging whether it is greater than or equal to 0, if yes: let

[0053] , and execute step 52) cyclically; if no, the iteration is ended, the optimal control variable sequence corresponding to the cumulative minimum operating cost is obtained, and the optimal control variable sequence is taken as the global optimal control trajectory of the hydraulic system. Step six: starting from the initial state, the globally optimal motor speed sequence and pump displacement sequence are reconstructed in the forward direction as the reference input of the control system, so as to realize the global energy consumption minimization operation of the hydraulic system.

[0054] Specifically, the optimal sequence is obtained by forward backtracking: the determination of the initial state follows the "target operating condition-minimum power" principle: under the given expected flow and the corresponding constraints (such as displacement / pressure) at a given time, based on the foregoing input power prediction model, the search space is taken as the allowable speed-displacement value range of the device, the current known operating condition parameters are combined for traversal solution, the speed and displacement that predict the minimum input power under the operating condition are selected as the system state at this time, and are taken as the starting node of dynamic programming, and the motor speed trajectory and pump displacement trajectory are gradually generated; inputting the trajectory into the power prediction model again, the optimized power curve can be obtained, and the total energy consumption of the hydraulic system under the optimized control is calculated by integration. In this way, the global energy consumption minimization control combining operating condition discretization and dynamic programming is realized.

[0055] Specifically, the optimal sequence is obtained by forward backtracking: the determination of the initial state follows the "target operating condition-minimum power" principle: under the given expected flow and the corresponding constraints (such as displacement / pressure) at a given time, based on the foregoing input power prediction model, the search space is taken as the allowable speed-displacement value range of the device, the current known operating condition parameters are combined for traversal solution, the speed and displacement that predict the minimum input power under the operating condition are selected as the system state at this time, and are taken as the starting node of dynamic programming, and the motor speed trajectory and pump displacement trajectory are gradually generated; inputting the trajectory into the power prediction model again, the optimized power curve can be obtained, and the total energy consumption of the hydraulic system under the optimized control is calculated by integration. In this way, the global energy consumption minimization control combining operating condition discretization and dynamic programming is realized.

[0056] ​​​Specifically, in this embodiment, the forward reconstruction step includes: starting from the initial state, gradually generating the optimal sequence of motor speed and pump displacement; inputting the sequence into the power prediction model, calculating the optimized total energy consumption, and outputting it as a control instruction.

[0057] The above embodiments are merely preferred embodiments for the purpose of fully illustrating the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are within the scope of protection of the present invention. The scope of protection of the present invention shall be subject to the claims.

Claims

1. A variable speed and variable displacement control method for an electro-hydraulic power source with global energy consumption minimization, characterized by: The steps include: Step 1: Obtain the operating data of the hydraulic system under multiple working conditions; Step 2: Build a system input power prediction model based on the extreme gradient boosting algorithm to learn the nonlinear mapping relationship between the operating data and the input power; Step 3: Discretize the continuously changing working conditions and divide them into multiple discrete time periods; Step 4: In each discrete time period, a set of candidate states of motor speed and pump displacement is generated, and the input power of each candidate state is predicted using the power prediction model; the generation of the candidate state set is limited to a dynamic change range determined by the maximum rate of change of motor speed and the maximum rate of change of pump displacement; Step 5: Based on the principle of dynamic programming, the minimum cumulative operating cost and the corresponding optimal control variables of each time period are calculated through reverse recursion to generate the global optimal speed sequence and displacement sequence that meet the dynamic constraints; Step 6: Starting from the initial state, forward reconstruct the globally optimal motor speed sequence and pump displacement sequence as the reference input of the control system to achieve global energy minimization operation of the hydraulic system.

2. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 1, characterized in that: In step 1, the method steps for obtaining operation data include: 11) Get the motor speed , outlet pressure , export traffic ,temperature and system input power , and based on the physical relationship of the pump port The pump displacement is derived; 12) Preprocess the running data, including removing outliers and normalizing them, and use a random partitioning strategy to divide them into training and test sets.

3. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 1, characterized in that: In step 2, the steps of constructing the input power prediction model include: 21) Build an input power prediction model using an extreme gradient boosting algorithm regression architecture, including: accelerating computation through a data matrix structure; utilizing a histogram tree construction method for efficient memory management; determining the optimal combination through hyperparameter grid search; and enabling multi-core parallel training to improve training efficiency. 22) Use the coefficient of determination And root mean square error RMSE evaluate model performance, including: determination coefficient is greater than a preset first threshold, and the root mean square error RMSE is less than a preset second threshold.

4. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 1, characterized in that: In step 3, the time step of the discretization process is The value range is 0.2-0.5 seconds to adapt to the dynamic response characteristics of pump displacement and motor speed and reduce sensitivity to noise.

5. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 1 is characterized in that: In step 4, the method for constructing the candidate state set includes: setting the dynamic change interval of the speed and displacement according to the dynamic adjustment capabilities of the motor and pump, and limiting the state jump between adjacent time periods; the dynamic change interval is expressed as: in: For the The state at a moment; and Respectively Motor speed and pump displacement at each moment; For the The state at a moment; and Respectively Motor speed and pump displacement at each moment; and Respectively represent The maximum speed change and displacement change allowed in a discrete time step; is the discrete grid set of motor speed; is the minimum motor speed allowed / controllable; is the maximum motor speed allowed / controllable; is the discrete grid set of pump displacement; is the minimum allowable / controllable pump displacement; is the maximum allowable / controllable pump displacement; and are the speed discrete grid step and the displacement discrete grid step respectively.

6. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 1, characterized in that: In step 5, the method steps for calculating the minimum cumulative operating cost and the corresponding optimal control amount in each time period by reverse recursion are as follows: 51) Order , is the number of discrete points, For the The number of operating points in a discrete time period, and ; 52) Initialize the candidate state set for the first operating point , from the candidate state set Select Status , the input power prediction model is used to predict the state Input power under ; 53) Calculation status Next The operating cost of a discrete time period , operating costs It is obtained by accumulating the energy consumption of two adjacent operating points in time steps, and for the Any two adjacent discrete time periods Operating point and Energy consumption of the operating point time step, , and its calculation steps are: 531) Status-based Generate the Candidate state set of operating point , from the candidate state set Select Status , the input power prediction model is used to predict the state Input power under ; Among them: if , then the state Candidate state set The currently selected state ;like , then the state Candidate state set Medium energy consumption Minimal state ; 532) Based on the adjacent Operating point and The input power of the operating point is used to determine the energy consumption of the time step of the two adjacent operating points by multiplying the average power by the time step. ; 533) Traverse the candidate state set , get in state Energy consumption Minimal state And the corresponding minimum energy consumption ; 54) Traverse the candidate state set , get the The operating cost is set in a discrete time period The smallest state combination of each operating point, and mapping the state combination to the corresponding control variable sequence; 55) Judgment Is it greater than or equal to 0? If so: , looping through step 52); if so, the iteration ends, and the optimal control variable sequence corresponding to the cumulative minimum operating cost is obtained, and the optimal control variable sequence is used as the global optimization control trajectory of the hydraulic system.

7. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 6, characterized in that: The control variable sequence includes a motor speed sequence and a pump displacement sequence.

8. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 6, characterized in that: The adjacent Operating point and The energy consumption at the operating point is: in: For the adjacent Operating point and The time step between load points.

9. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 6, characterized in that: In the step 53), The cumulative minimum operating cost at the operating point is expressed by the cumulative energy consumption cost function: in: Indicates the discrete time periods in the The working point is always in the state The minimum cumulative operating cost when For the discrete time periods in the Optimal cumulative energy consumption at the operating point; Status Transfer to state The corresponding energy consumption cost; A pair of adjacent states.

10. The method for controlling a variable speed and displacement electro-hydraulic power source with global energy consumption minimization according to claim 1, characterized in that: In step 6, the forward reconstruction step includes: Starting from the initial state, the optimal sequence of motor speed and pump displacement is gradually generated; The sequence is input into a power prediction model, the optimized total energy consumption is calculated, and the optimized total energy consumption is output as a control instruction.

Images