A method for engine speed compensation in a hybrid excavator and a hybrid excavator

Abstract

This invention discloses a method for compensating engine speed in a hybrid excavator and a hybrid excavator. The method sets a fuzzy subset based on the required torque, current SOC value, and the difference between engine speed and motor speed. Through a fuzzy control algorithm, the ISG motor torque is dynamically adjusted. Since the fuzzy control uses fixed fuzzy rules, a Q-table is established using a Q-learning algorithm to obtain the ISG motor compensation torque. The final output ISG motor torque is determined to be the sum of the fuzzy controller output torque and the compensation torque output by the Q-learning algorithm, thus determining the ISG motor output torque. This method leverages the fast response speed of the motor torque to effectively compensate for the difference between the required torque and the actual torque, avoiding engine speed fluctuations caused by excessive engine torque changes, and improving fuel economy and driving stability.

Description

Technical Field

[0001] This invention relates to the field of power compensation technology for engineering machinery, specifically to a method for engine speed compensation in a hybrid excavator and a hybrid excavator. Background Technology

[0002] As a sector heavily reliant on energy consumption and environmental pollution, the development of energy-saving technologies in the construction machinery industry has always faced severe challenges. Hydraulic excavators, as typical high-energy-consuming off-road mobile machinery, suffer from inefficient engine operation due to frequent and drastic load fluctuations, resulting in generally low fuel efficiency. In traditional hydraulic excavator systems, the engine directly drives the hydraulic pump as the power source. When faced with complex and variable working conditions, the power mismatch between the power source and the load is particularly pronounced, leading to significant fluctuations in the engine's operating point and preventing stable operation within the efficient fuel-consuming range.

[0003] Hybrid excavators, compared to traditional fuel-powered excavators, utilize a hybrid drive system, employing both an engine and an electric motor as power sources. The range of pure electric vehicles is a key concern, and hybrid systems effectively address this issue. To further optimize the smoothness and stability of hybrid excavators and mitigate the impact of engine speed fluctuations on driving experience, motor-assisted engine speed compensation has become an important technical approach. Speed ​​compensation refers to the ability of the excavator to quickly adjust its speed during operation due to the rapid response of the electric motor, thus reducing vehicle vibration caused by engine speed drops.

[0004] Chinese invention patent publication number CN117702855A discloses a control system and method applicable to excavator engines and ISG motors. It employs a mechanism where feedforward torque reduces speed to zero at a preset slope. However, this preset slope relies on fixed parameters, making it difficult to dynamically adapt to different operating conditions (such as cold starts and high-load shutdowns). When the actual friction torque or battery status deviates from the calibrated range, it easily leads to speed tracking deviations, requiring an additional feedback correction stage.

[0005] Chinese invention patent publication number CN120026996A discloses an engine shutdown method, device, equipment, and storage medium. This solution requires simultaneous processing of multiple source signals such as hydraulic pump pressure difference, displacement difference, and engine power. The calculation of load feedforward compensation has a millisecond-level delay. Under rapid transient conditions such as sudden unloading of excavators, the compensation lag may lead to a delay in engine speed recovery. Summary of the Invention

[0006] To address the issue of engine speed fluctuation in hybrid excavators, this invention proposes a method for compensating engine speed in hybrid excavators. This method establishes a fuzzy subset based on the required torque, current SOC value, and the difference between engine speed and motor speed. Through a fuzzy control algorithm, the ISG motor torque is dynamically adjusted to determine the magnitude of the motor's output torque. This method leverages the fast torque response of the motor to effectively compensate for the difference between the required torque and the actual torque, avoiding engine speed fluctuations caused by excessive changes in engine torque, thus improving fuel economy and driving stability.

[0007] To achieve the above objectives, the present invention employs the following technical solution:

[0008] This invention provides a method for compensating the engine speed of a hybrid excavator, comprising:

[0009] Collect engine data of the hybrid excavator under current operating conditions, including the vehicle's required torque, battery SOC value, engine speed, and ISG motor speed;

[0010] A fuzzy controller is established, and the fuzzy subset of the input variables, the output quantity, and the fuzzy rules of the fuzzy controller are determined. The input variables are the vehicle's required torque, the battery's SOC value, and the difference between the engine speed and the ISG motor speed. The output quantity is the ideal ISG motor output torque.

[0011] Torque compensation is performed based on the Q-learning algorithm to obtain the compensation value;

[0012] The ideal ISG motor output torque output by the fuzzy controller is superimposed with the compensation value to form the final ISG motor output torque.

[0013] Preferably, the operating conditions include:

[0014] Sudden increases or decreases in load, low SOC, and power and speed close to target values ​​require limiting power and adjusting speed accordingly.

[0015] Preferably, the collection of hybrid excavator engine data under current operating conditions includes:

[0016] In the current working cycle of lowering-digging-lifting-rotating-shovel lowering-returning, 10 sets of low SOC, rock conditions; 10 sets of high SOC, rock conditions; 10 sets of low SOC, soil conditions; and 10 sets of high SOC, soil conditions were excavated; a total of 40 sets of cycle working condition data were collected.

[0017] Preferably, the fuzzy subset representation of the input variables is as follows:

[0018] Demand torque The fuzzy subset is {NB, NM, NS, ZE, PS, PM, PB};

[0019] Battery SOC value The fuzzy subset is {NB, NM, NS, ZE, PS, PM, PB};

[0020] The difference between engine speed and ISG motor speed The fuzzy subset is {NB, NM, NS, ZE, PS, PM, PB}.

[0021] Preferably, the determination of the fuzzy subset of the input variables of the fuzzy controller further includes: The quantization domain for determining the fuzzy subset of the demand torque is {-100, 100}; The quantization universe of discourse for determining the fuzzy subset of the battery SOC value is {0, 100}; The quantization domain for determining the difference between engine speed and ISG motor speed is {-300, 300}.

[0022] Preferably, the fuzzy rules are set as follows:

[0023] The fuzzy rule under sudden load increase conditions is: if the required torque = Positive battery SOC value =zero and the difference between engine speed and ISG motor speed =Negative, then the output torque = Center;

[0024] The fuzzy rule under the condition of sudden load reduction is: if the required torque =Negative and battery SOC value =zero and the difference between engine speed and ISG motor speed = at the center, then the output torque =Negative middle;

[0025] The fuzzy rule for power limiting conditions when SOC is too low is: if the required torque... = Positive battery SOC value =The negative value and the difference between the engine speed and the ISG motor speed =If the negative value is large, the output torque will be large. =positive small;

[0026] The fuzzy rule for operating conditions where the rotational speed is close to the target value is: if the required torque =Zero and battery SOC value =zero and the difference between engine speed and ISG motor speed =zero, then the output torque =Zero.

[0027] Preferably, the establishment of the fuzzy controller further includes:

[0028] Select appropriate membership functions for the fuzzy language. The membership function type is Trimf trigonometric function, and the number of membership functions is 7.

[0029] Preferably, the torque compensation based on the Q-learning algorithm to obtain the compensation value includes:

[0030] The state of the Q-learning algorithm is defined as the error between the Q-value of the current iteration and the updated Q-value. and error change rate ;

[0031] The action of the Q-learning algorithm is defined as the compensation value of the fuzzy controller output. ;

[0032] The reward for the Q-learning algorithm is defined as: ;in As a reward, , , These are the weighting coefficients. This is the compensation value;

[0033] Initialize the Q-table, load the Q-learning algorithm, and continue until convergence to obtain the optimal compensation value.

[0034] Preferably, in the Q-learning algorithm process, the following is adopted: Strategy selection action.

[0035] The present invention also provides a hybrid excavator, which uses the above-mentioned hybrid excavator engine speed compensation method to obtain the output torque of the ISG motor.

[0036] The beneficial effects of the technical solution of this invention are as follows:

[0037] (1) This invention proposes a hybrid excavator engine speed compensation strategy. A fuzzy subset is set based on the required torque, current SOC value, and the difference between engine speed and motor speed. The ISG motor torque is dynamically adjusted using a fuzzy control algorithm. Since the fuzzy control uses fixed fuzzy rules, a Q-table is established using a Q-learning algorithm to obtain the ISG motor compensation torque. The final output ISG motor torque is determined to be the sum of the fuzzy controller output torque and the compensation torque output by the Q-learning algorithm, thus determining the magnitude of the motor output torque. This method utilizes the advantage of fast motor torque response speed to effectively compensate for the difference between the required torque and the actual torque, avoiding engine speed fluctuations caused by excessive engine torque changes, and improving fuel economy and driving operation stability.

[0038] (2) This invention addresses the issue of traditional fuzzy rule bases which rely on experience to set fixed rules. To address various complex operating conditions, a Q-learning algorithm is proposed. First, a fuzzy controller is used to generate basic control quantities. Then, a compensation quantity is dynamically generated using a Q-learning algorithm in discrete state space. The two are superimposed to form the final control command. An innovative improved ε-greedy exploration strategy and reward function setting are introduced to achieve adaptive improvement in control accuracy while preserving the robustness of fuzzy control. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the control principle of the hybrid excavator engine speed compensation method provided by the present invention;

[0040] Figure 2 This is a schematic diagram illustrating the compensation principle of the Q-learning algorithm provided by the present invention;

[0041] Figure 3 This is a schematic diagram of the compensation process for the Q-learning algorithm provided by the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0043] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0044] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0045] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0046] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0047] It should be emphasized here that the step markers mentioned below are not a limitation on the order of the steps, but should be understood as meaning that the steps can be executed in the order mentioned in the embodiments, or in a different order than in the embodiments, or several steps can be executed simultaneously.

[0048] Firstly, this invention provides a method for compensating the engine speed of a hybrid excavator, applicable to hybrid excavators. Its principle is as follows: Figure 1 As shown, the method includes the following steps:

[0049] S1. Operating Data Acquisition and Analysis

[0050] Hybrid hydraulic excavators exhibit significant periodicity and load fluctuation characteristics in their operating conditions. A typical work cycle is as follows: lowering - digging - lifting - rotating - releasing the shovel - returning to position, with the entire process taking about 15 seconds. During the entire continuous process, the load demand changes drastically.

[0051] In this invention, data such as the required torque of the vehicle, the battery SOC value, the engine speed, and the ISG motor speed are collected by installing a data acquisition device.

[0052] Based on the above typical excavator operating cycle, it can be divided into several typical operating conditions such as sudden load increase, sudden load decrease, low SOC requiring power limitation, and speed approaching the target value.

[0053] For example, for the typical working conditions mentioned above, different sites need to be selected during the data collection process. This could involve excavating soil or rocks, while also considering the battery's SOC (State of Charge). Both low and high SOCs need to be tested. In summary, during the work cycle of lowering, excavating, lifting, rotating, lowering the shovel, and returning to position, 10 sets of low SOC, rock conditions; 10 sets of high SOC, rock conditions; 10 sets of low SOC, soil conditions; and 10 sets of high SOC, soil conditions were collected, for a total of 40 sets of cyclical working condition data.

[0054] S2. Establish the Fuzzy controller structure

[0055] Hybrid excavators consist of an engine and an ISG motor. By reducing engine speed fluctuations and keeping the engine speed within its optimal operating range, fuel efficiency is achieved. This is achieved by determining the input and output quantities based on the specific system; here, a standard three-dimensional control structure is chosen, where the input is the required torque. ,Battery The difference between engine speed and ISG motor speed The corresponding fuzzy quantity of the fuzzy controller is , and The output is the ISG motor compensation torque value. The specific description is as follows:

[0056] S21. Input variables (fuzzy subset)

[0057] Demand torque The total power demand of the power system under the current operating conditions is the main cause of speed fluctuations. The fuzzy subset of the language describing the input variables is determined as {NB, NM, NS, ZE, PS, PM, PB}. Next, the universe of discourse for the input variables is set, namely the demand torque. The quantization domain is {-100, 100}.

[0058] Current SOC value This reflects the available energy state of the power battery and is a key factor constraining motor enablement / recharge (too low SOC requires limiting discharge, too high requires limiting charging). The fuzzy subset of the language describing the input variables is determined to be {NB, NM, NS, ZE, PS, PM, PB}. Next, the universe of discourse for the input variables is set, i.e., the current SOC value. The quantization domain is {0, 100}.

[0059] The difference between engine speed and ISG motor speed This directly reflects the speed fluctuation trend at the power coupling point; a large difference may indicate an impending speed fluctuation. The fuzzy subset describing the input variable language is determined to be {NB, NM, NS, ZE, PS, PM, PB}. Next, the universe of discourse for the input variable is set, namely the difference between the engine speed and the ISG motor speed. The quantization domain is {-300, 300}.

[0060] By demanding torque Current SOC value The difference between engine speed and ISG motor speed By dividing the universe of discourse, the required torque can be achieved. Current SOC value The difference between engine speed and ISG motor speed The division of different working ranges allows for dynamic adjustment of the ISG motor output torque via a fuzzy controller, achieving the function of compensating for engine speed. Finally, appropriate membership functions are selected for the fuzzy language; the membership function type is Trimf (trigonometric function), and the number of membership functions is 7. These membership functions are then modified according to the requirements.

[0061] S22. Design of Fuzzy Reasoning Decision Algorithm

[0062] To perform fuzzy inference based on fuzzy control rules and determine the fuzzy output, the fuzzy rules must first be determined.

[0063] Based on the above typical working conditions, the following design principles for fuzzy rules apply:

[0064] When the difference between the engine speed and the ISG motor speed When the value is negative, that is, when the speed is lower than the target value, it is positive compensation; otherwise, it is negative compensation.

[0065] When the current SOC value When the battery is in a state of negative large or negative small, it is in a low charge area. At this time, the compensation should be reduced to prioritize the protection of battery life.

[0066] When demand torque The greater the rate of torque change, the more aggressive the torque compensation strategy should be, and the faster the motor speed will climb.

[0067] Examples of fuzzy rules corresponding to different working conditions are shown in Table 1 below:

[0068] Table 1 Examples of Fuzzy Rule Bases

[0069]

[0070] By defuzzifying the results, the ideal output torque of the ISG motor can be obtained.

[0071] S3. Establishment of a self-learning fuzzy rule base

[0072] Traditional fuzzy rule bases set fixed rules based on experience. When the SOC power drops suddenly or the engine speed increases suddenly, the fixed rules cannot adjust the compensation intensity in real time, making it difficult to adapt to complex scenarios such as sudden load changes and multiple working condition switching of hydraulic excavators.

[0073] When the excavator is digging through hard rock, the torque demand surges, putting the excavator under overload and impact conditions. When the battery's state of equilibrium (SOH) is below 80%, battery capacity decays, potentially causing a sudden change in battery SOC, indicating the battery is in an aging state. When the excavator is in extremely cold conditions, changes in hydraulic oil viscosity lead to sluggish vehicle movement. Based on this, this invention provides a torque compensation rule based on the Q-learning algorithm, see [link to relevant documentation]. Figure 1 The specific implementation process is as follows:

[0074] S31, Define the state.

[0075] The state should adequately represent the current system condition and control requirements. The state space needs to be discretized or effectively represented by a Q-function. In this invention, the state is defined as: the error between the Q-value of the current iteration and the updated Q-value in the Q-learning algorithm, expressed as: and the rate of change of error .

[0076] S32, Define the action.

[0077] After the fuzzy controller outputs the ideal torque of the ISG motor, the output value of the fuzzy controller needs to be dynamically corrected. The correction amount is... In this invention, actions are defined. This is the correction amount.

[0078] The action space is then discretized, with a discretization range of {-0.5, -0.3, -0.1, 0, +0.1, +0.3, +0.5}; the final control output value is:

[0079] ,

[0080] in This is the final output after adjustment using the Q algorithm. The output of the fuzzy controller.

[0081] S33, Define Rewards

[0082] The reward signal is the primary driver of the Q-learning algorithm and directly reflects the quality of the current control performance. A common design is as follows:

[0083] ,

[0084] ,

[0085] - ,

[0086] in , These are the weighting coefficients. The error of the input variable for the fuzzy controller. The value is the rate of change of error, the negative sign indicates a penalty, and the overall objective is to minimize the value within the parentheses.

[0087] Combining the above common designs, the reward function for multi-objective purposes is as follows:

[0088] .

[0089] Safety constraints are implemented by constructing a reward function. By giving a negative reward (penalty), a large negative reward is given when a dangerous action is taken or a dangerous state is entered. This will cause the Q value to decrease, thereby achieving the function of safety constraints.

[0090] S34. Initialize the Q table.

[0091] Create a table Q( ), where each row corresponds to all possible states. The column corresponds to all possible actions. The initial value is usually set to 0 or a small random number.

[0092] S35. Establish the Q learning process, such as Figure 3 As shown:

[0093] S351. Observe the current state. : and .

[0094] S352, Select Action :use Strategy, with probability 1- Select current status The specific action that maximizes Q value, With probability Choose an action randomly.

[0095] S353, Execution Action: The final output control quantity is: .

[0096] S354, Observe the next state and rewards After the system runs one step, based on , , The system responds to measure the new state and calculates the reward.

[0097] S355, Update Q table: Use Q to learn update rules:

[0098] ,

[0099] in The learning rate controls the speed at which new information overwrites old information. ; This is a discount factor used to measure the importance of future rewards. .

[0100] S356, State Transition: .

[0101] S36. Convergence: As the number of learning steps increases, the Q-table will gradually converge when sufficient exploration and utilization are achieved; when the condition is met... At that time, the optimal compensation amount is obtained. .

[0102] This invention employs the aforementioned technical solution. First, a fuzzy controller is used to generate basic control quantities. Then, a compensation quantity is dynamically generated using a Q-learning algorithm in discrete state space. The two are then superimposed to form the final control command. An innovative improved ε-greedy exploration strategy and safety constraint mechanism are introduced, achieving adaptive improvement in control accuracy while preserving the robustness of fuzzy control.

[0103] This invention employs a fuzzy rule-based control algorithm. The fuzzy controller controls the torque of the ISG motor to ensure it outputs an ideal working torque, effectively compensating for the difference between the required torque and the actual torque, and preventing engine speed fluctuations caused by excessive engine torque variations. This method solves the problem in hybrid excavators where changes in load demand, such as boom lifting and slewing start-stop, cause the engine output torque to lag behind the required torque, leading to engine speed fluctuations.

[0104] It should be noted that the ISG motor torque regulation method in this invention is based on fuzzy control algorithm for prediction. However, different algorithms such as particle swarm optimization and the cuckoo algorithm can also be selected for ISG motor torque regulation. In the process of fuzzy rule self-learning, traditional fuzzy control uses fixed rules, which cannot respond promptly to complex operating conditions. The Q-learning algorithm is used to dynamically adjust the ISG motor torque output by the fuzzy control. Other reinforcement learning algorithms, such as SARSA and DDPG, can also be used here. This invention only provides a basic description of the implementation process. Any simple modifications, equivalent changes, or alterations made to the above process based on the technical essence of this invention shall still fall within the scope of this technology.

Claims

1. A hybrid shovel engine rotation speed compensation method characterized by comprising: include: Collect engine data of the hybrid excavator under the current working conditions, including the vehicle's required torque, battery SOC value, engine speed, and ISG motor speed; the working conditions include: sudden load increase, sudden load decrease, low SOC requiring power limiting, and speed approaching the target value; A fuzzy controller is established, and the fuzzy subset of its input variables, its output, and the fuzzy rules are determined. The input variables are the vehicle's required torque, the battery's SOC value, and the difference between the engine speed and the ISG motor speed. The output is the ideal ISG motor output torque. The fuzzy rules are set as follows: The fuzzy rule under the load surge condition is: if the demand torque = positive large and the battery SOC value = zero and the difference between the engine speed and the ISG motor speed = negative medium, then the output torque = positive medium; The fuzzy rule under the condition of sudden load reduction is: if the required torque =Negative and battery SOC value =zero and the difference between engine speed and ISG motor speed = at the center, then the output torque =Negative middle; The fuzzy rule for power limiting conditions when SOC is too low is: if the required torque... = Positive battery SOC value =The negative value and the difference between the engine speed and the ISG motor speed =If the negative value is large, the output torque will be large. =positive small; The fuzzy rule for operating conditions where the rotational speed is close to the target value is: if the required torque =Zero and battery SOC value =zero and the difference between engine speed and ISG motor speed =zero, then the output torque =zero; Torque compensation is performed based on the Q-learning algorithm to obtain the compensation value; The ideal ISG motor output torque output by the fuzzy controller is superimposed with the compensation value to form the final ISG motor output torque.

2. The method for compensating the engine speed of a hybrid excavator according to claim 1, characterized in that, The collection of hybrid excavator engine data under current operating conditions includes: In the current working cycle of lowering-digging-lifting-rotating-shovel lowering-returning, 10 sets of low SOC, rock conditions; 10 sets of high SOC, rock conditions; 10 sets of low SOC, soil conditions; and 10 sets of high SOC, soil conditions were excavated; a total of 40 sets of cycle working condition data were collected.

3. The method for compensating the engine speed of a hybrid excavator according to claim 1, characterized in that, The fuzzy subset representation of the input variables is as follows: Demand torque The fuzzy subset is {NB, NM, NS, ZE, PS, PM, PB}. Battery SOC value The fuzzy subset is {NB, NM, NS, ZE, PS, PM, PB}. The difference between engine speed and ISG motor speed The fuzzy subset is {NB, NM, NS, ZE, PS, PM, PB}.

4. The method for compensating the engine speed of a hybrid excavator according to claim 3, characterized in that, The determination of the fuzzy subset of the input variables of the fuzzy controller also includes: The quantization domain for determining the fuzzy subset of the demand torque is {-100, 100}; The quantization universe of discourse for determining the fuzzy subset of the battery SOC value is {0, 100}; The quantization domain for determining the difference between engine speed and ISG motor speed is {-300, 300}.

5. The method for compensating the engine speed of a hybrid excavator according to claim 1, characterized in that, The establishment of the fuzzy controller also includes: Select appropriate membership functions for the fuzzy language. The membership function type is Trimf trigonometric function, and the number of membership functions is 7.

6. The method for compensating the engine speed of a hybrid excavator according to claim 1, characterized in that, The torque compensation based on the Q-learning algorithm, to obtain the compensation value, includes: The state of the Q-learning algorithm is defined as the error between the Q-value of the current iteration and the updated Q-value. and error change rate ; The action of the Q-learning algorithm is defined as the compensation value of the fuzzy controller output. ; The reward for the Q-learning algorithm is defined as: ;in As a reward, , , These are the weighting coefficients. This is the compensation value; Initialize the Q-table, load the Q-learning algorithm, and continue until convergence to obtain the optimal compensation value.

7. The method for compensating the engine speed of a hybrid excavator according to claim 6, characterized in that, In the Q-learning algorithm process, the following is adopted: Strategy selection action.

8. A hybrid excavator, characterized in that, The ISG motor output torque is obtained by using the hybrid excavator engine speed compensation method according to any one of claims 1 to 7.

Images