A diesel engine rotating speed active disturbance rejection control method based on nonlinear scaling

Abstract

The application discloses a diesel engine speed self-disturbance control method based on nonlinear scaling, which comprises the following steps: step 1, constructing a diesel engine speed dynamic model; the diesel engine speed dynamic model comprises total disturbance f(x, t); step 2, constructing a nonlinear scaling function g(z) to scale and reconstruct a tracking error e x ; step 3, based on the reconstructed variable x m of step 2, constructing an extended state observer of the lifting order, and feeding back an estimated value of the total disturbance f(x, t) to the diesel engine speed dynamic model of step 1 as the total disturbance f(x, t); step 4, constructing a low-gain control law, outputting u(t), and feeding back to the diesel engine speed dynamic model of step 1 to form a closed-loop system. The application can effectively inhibit high-frequency noise disturbance and reduce control input fluctuation.

Description

Technical Field

[0001] This invention relates to the field of diesel engine electronic control technology, and in particular to a diesel engine speed active disturbance rejection control method based on nonlinear scaling, which is especially suitable for precise control of diesel engine speed. Background Technology

[0002] Diesel engines are a crucial research area in the machinery industry. Due to their advantages such as high thermal efficiency, large torque, good economy, wide power range, rapid start-up, easy maintenance, safe operation, and long service life, they are widely used in non-road generator sets, on-road vehicles, ships, agricultural machinery, and construction machinery, serving as the primary power source for these devices. The speed control performance of diesel generators directly affects the system's operational stability and efficiency. However, as a typical nonlinear power system, diesel engines face numerous challenges in control, including the difficulty in measuring load torque and the significant high-frequency noise often present in speed measurements.

[0003] In power generation applications, diesel generator sets need to cope with complex and ever-changing electrical loads. When external loads suddenly change or random loads fluctuate abruptly, traditional control methods typically rely passively on speed deviations for adjustment, resulting in sluggish speed response and difficulty in meeting the requirements for fast and precise control. Furthermore, traditional speed control schemes usually require staged calibration of control parameters for steady-state and transient conditions, which involves a large calibration workload and poor parameter versatility. The same set of control parameters may trigger over-response at low speeds, even leading to engine surging.

[0004] Currently, proportional-integral-derivative (PI-DE) controllers are widely used in engineering control due to their simplicity and reliability, playing a crucial role in speed control problems. For example, in the research on "Intelligent robust PI adaptive control strategy for speed control of EV(s)," the authors used the nonlinear least squares method to obtain the variable parameters of the controlled object online and combined them with a PI controller to achieve the ability to adapt to changes in model parameters. Another example is the invention "A speed control strategy for sudden load changes in a generator diesel engine based on PID control" (application number CN202311437487.5). This invention compares the sum of the control outputs of a PID controller with a set threshold, and combines this with the positive or negative sign of the speed deviation and the rate of change of speed to switch between freezing and non-freezing modes for the integral I term, effectively reducing speed overshoot after sudden unloading and shortening the speed recovery time.

[0005] However, PID controller is based on linear feedback, which fails to utilize the system's dynamic model information, thus limiting the control accuracy when dealing with nonlinear systems such as diesel engines. Meanwhile, to improve disturbance rejection capability, control gain is usually increased, which often leads to severe fluctuations in control input under steady-state conditions, resulting in increased energy consumption and intensified noise problems.

[0006] For the problem of speed control, in addition to PID controllers, adaptive control, model predictive control (MPC), neural network control, and active disturbance rejection control methods have also been widely studied. In the research work of Fast Nonlinear Model Predictive Control on FPGA Using Particle Swarm Optimization, the authors used a nonlinear MPC algorithm to solve the problem of nonlinear disturbance in the diesel engine speed model, and the experimental results showed that this method performed well in transient tracking and disturbance suppression. For example, the application number CN201910045064.6 "A BP neural network combustion-speed double-loop control method for marine diesel engines", the invention provides a BP neural network combustion-speed double-loop control algorithm for diesel engines, which reduces fuel consumption and reduces noise. However, due to the limitations of embedded control system computing resources, real-time operation of the MPC algorithm still faces certain challenges, and the model size of the BP neural network is also limited, further limiting the control accuracy.

[0007] Active disturbance rejection control (ADRC) estimates the total disturbance of the system (including internal unmodeled dynamics and external disturbances) in real time through an extended state observer (ESO), and exhibits good robustness, making it a potential solution for diesel engine speed control. In the research work of New design of active disturbance rejection control for nonlinear uncertain systems with unknown control input gain, the authors designed a new active disturbance rejection control (ADRC) method for nonlinear uncertain systems with unknown control input gain. This method only relies on the information of the control direction, i.e., only the positive and negative information of the control gain is needed, without the need to determine the specific value of the control gain, thus effectively dealing with the major uncertainty problem caused by unknown input gain. For example, the application number CN202210550397.6 "A diesel engine speed cascade active disturbance rejection control system, diesel engine and locomotive", the invention focuses on introducing a diesel engine speed cascade active disturbance rejection control system, which includes a feedforward controller, an inner loop active disturbance rejection controller, and an outer loop active disturbance rejection controller, providing a basic paradigm for the design of active disturbance rejection controllers for diesel engine speed problems.

[0008] However, the existing ADRC controller generally uses fixed control parameters, resulting in a significant trade-off between disturbance rejection capability and control input fluctuation, and it is difficult to balance both. Therefore, it is urgent to develop a new diesel engine speed control method to enhance the disturbance rejection capability while effectively reducing the control input fluctuation, while meeting the real-time operation requirements, so as to comprehensively improve the operation performance and economy of the diesel engine. SUMMARY

[0009] The purpose of the present application is to address the significant limitations of the conventional diesel engine speed control method in suppressing disturbances and reducing control input fluctuations in the prior art. Conventional control methods such as PID and MPC cannot balance the contradiction between disturbance suppression capability and control input fluctuation. In addition, the existing ADRC method still faces the trade-off problem between disturbance suppression and control input fluctuation under constant parameters, and a diesel engine speed self-disturbance rejection control method based on nonlinear scaling is provided.

[0010] The technical solution adopted to achieve the purpose of the present application is:

[0011] A diesel engine speed self-disturbance rejection control method based on nonlinear scaling, comprising the following steps:

[0012] Step 1, constructing a diesel engine speed dynamic model: the controlled output of the diesel engine speed dynamic model is x(t), x(t)=n(t), n(t) is the diesel engine speed, the controlled input is u(t), u(t)=M i (t), M i (t) is the indicated torque of the diesel engine, and the diesel engine speed dynamic model includes total disturbance f(x,t);

[0013] Step 2, constructing a nonlinear scaling function g(z) to scale and reconstruct the tracking error e x , e x =x(t)-r(t), r(t) is the reference diesel engine speed, and the reconstructed state variable x m =g(z)e x +r(t);

[0014] Step 3, constructing a lifting order extended state observer based on the reconstructed variable x m from step 2, outputting an estimated value of the total disturbance f(x,t) , which is fed back to the diesel engine speed dynamic model in step 1 as the total disturbance f(x,t);

[0015] Step 4, based on the nonlinear scaling function of step 2 and the estimated value of the total disturbance f(x,t) obtained in step 3 A low-gain control law is constructed to output u(t) and fed back to the diesel engine speed dynamic model in step 1 to form a closed-loop system.

[0016] In the technical solution, the diesel engine speed dynamic model in step 1 is:

[0017]

[0018] wherein x(t) is the controlled output, x(t)=n(t), n(t) is the diesel engine speed; u(t) is the control input, b0 is the control gain, J is the moment of inertia; m1, m2 and m3 are three coefficients; f(x,t) is the total disturbance.

[0019] In the technical solution, the nonlinear scaling function in step 2 is:

[0020]

[0021] wherein, E is the allowable speed error fluctuation range, and c is a parameter to be calibrated.

[0022] In the technical solution, in step 3, the rising and falling order extended state observer is:

[0023]

[0024] wherein β1 and β2 are observer bandwidth-related parameters, is the total disturbance estimated by the rising and falling order extended state observer, is the estimated value of the first derivative of the total disturbance f(x,t), and ξ1 and ξ2 are intermediate variables.

[0025] In the technical solution, β1=ω0 2 , β2=2ω o , ω o represents the bandwidth of the rising and falling order extended state observer.

[0026] In the technical solution, in step 4, the low-gain control law is:

[0027]

[0028] obtained from step 3, k p is the gain of the proportional controller in the low-gain control law.

[0029] In the technical solution, k p =ω c , ω cis the pole position of the closed-loop system after pole placement.

[0030] Compared with the prior art, the present application has the beneficial effects that:

[0031] 1. Introduction of nonlinear scaling function: The present application first introduces a continuous and derivable nonlinear scaling function in the diesel engine speed control, effectively suppresses high-frequency noise interference and reduces control input fluctuations by dynamically adjusting the feedback control gain. Compared with the traditional linear feedback mechanism, the nonlinear scaling function can adaptively adjust the scaling amplitude according to the error size, thereby achieving a balance between steady-state control accuracy and dynamic response performance.

[0032] 2. Design of diesel engine speed self-disturbance rejection control (NSADRC) based on nonlinear scaling: The total disturbance and its first derivative are accurately estimated by the nonlinear scaling function-based reduced-order extended state observer (SARESO), and the state variable scaled by the scaling function is introduced, further reducing the sensitivity of the observer to noise. Compared with the traditional full-order or reduced-order extended state observer, the SARESO designed in the present application improves the estimation accuracy and accuracy while reducing the high-frequency noise in the total disturbance estimate under steady-state conditions, ultimately reducing the fluctuations of the control input.

[0033] 3. Application of low-gain control strategy: The present application proposes a low-gain control law that dynamically adjusts the control input amplitude by combining the disturbance estimation results of the nonlinear scaling function and the reduced-order extended state observer, effectively reducing the oscillation of the steady-state control input while maintaining excellent disturbance rejection performance under transient conditions. This strategy breaks through the trade-off between disturbance suppression and control input fluctuations in traditional control methods. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 The figure shows the overall block diagram of the present application.

[0035] Figure 2 The figure shows the nonlinear scaling function and its first derivative. DETAILED DESCRIPTION

[0036] The present application will be further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0037] Example 1

[0038] As Figure 1 shown, a diesel engine speed self-disturbance rejection control method based on nonlinear scaling includes the following steps:

[0039] Step 1: Construct a dynamic model of diesel engine speed. The controlled output of the dynamic model is the diesel engine speed n(t), and the controlled input is u(t), u(t) = M. i (t), M i (t) represents the indicated torque of the diesel engine; the dynamic model of the diesel engine speed includes a total disturbance f(x,t);

[0040] Step 2, construct a nonlinear scaling function g(z) to account for the tracking error e. x Perform scaling and reconstruction, e x = x(t) - r(t), where r(t) is the reference diesel engine speed, and the reconstructed state variable x m =g(z)e x +r(t);

[0041] Step 3, based on the reconstructed variable x from Step 2 m Construct an ascending-descending extended state observer to output an estimate of the total perturbation f(x,t). The dynamic model of diesel engine speed fed back from step 1 is used as the total disturbance f(x,t):

[0042] Step 4: Based on the nonlinear scaling function from Step 2 and the estimate of the total perturbation f(x,t) from Step 3. A low-gain control law is constructed to output u(t), which is then fed back to the dynamic model of the diesel engine speed in step 1 to form a closed-loop system.

[0043] Example 2

[0044] like Figure 1 As shown, a diesel engine speed active disturbance rejection control method based on nonlinear scaling includes the following steps:

[0045] Step 1, construct a dynamic model of diesel engine speed:

[0046] The dynamic model of diesel engine speed includes unmodeled dynamic disturbances and external interferences. Specifically, the dynamic model of diesel engine speed is represented as follows:

[0047]

[0048] Where J is the moment of inertia, representing the inertia of the crankshaft and the rigidly connected shaft system, and M... i (t) represents the indicated torque of the diesel engine, M fri (n,t) represents the internal friction torque of the diesel engine, M load n(t) represents the external load torque, n(t) represents the diesel engine speed, and t represents time. The first derivative representing the diesel engine speed, i.e. the rate of change of the diesel engine speed, is determined by the moment of inertia, the diesel engine indicated torque, the friction torque, and the external load torque.

[0049] Mfriction(n, t) is the friction torque of the diesel engine fri (n, t) is the friction torque model, which is fitted by using the experimental data with the nonlinear least square method. The friction torque is:

[0050] Mfriction(n, t) = Mfriction(n) + Mfriction(n, t) fri (n, t) = Mfriction(n) + Mfriction(n, t) kn (n, t) = Mfriction(n) + Mfriction(n, t) u (n, t) = Mfriction(n) + Mfriction(n, t)

[0051] Mfriction(n, t) = Mfriction(n) + Mfriction(n, t) kn Mfriction(n) is the quadratic polynomial fitting model of the known part of the friction torque, and Mfriction(n, t) is the deviation part of the friction torque, which represents the difference between the experimental measured value and the fitted value. u Mfriction(n, t) = Mfriction(n) + Mfriction(n, t) kn Mfriction(n, t) = Mfriction(n) + Mfriction(n, t)

[0052] Mfriction(n, t) = Mfriction(n) + Mfriction(n, t) kn Mfriction(n, t) = Mfriction(n) + Mfriction(n, t) 2

[0053] Mfriction(n, t) = Mfriction(n) + Mfriction(n, t)

[0054] Mfriction(n, t) = Mfriction(n) + Mfriction(n, t) fri Mfriction(n, t) = Mfriction(n) + Mfriction(n, t)

[0055]

[0056] The control tasks of the diesel engine speed control mainly include the following targets: (1) speed tracking: design appropriate diesel engine indicated torque M i (t) to make the diesel engine speed n(t) accurately track the predetermined reference signal r(t), i.e. the reference diesel engine speed; (2) fluctuation suppression: in steady state operation, try to reduce the fluctuation amplitude of the control input M i (t); (3) disturbance rejection performance: in the case of sudden change of external load torque M load (t) and existence of internal model parameter M u (n, t) uncertainty, try to reduce the deviation and steady state fluctuation of the diesel engine speed n(t). To improve the processing capability of unmodeled dynamic disturbance and external disturbance, the diesel engine dynamics model is expanded to the form containing total disturbance f(x, t):

[0057]

[0058] where the controlled output x(t) is the diesel engine speed n(t), i.e. x(t) = n(t), and the control input u(t) of the system is the diesel engine indicated torque M i(t), that is, u(t) = M i (t), control gain Dynamic model of the known part The total unknown disturbance, which consists of the disturbances unmodeled inside the system and the difficult-to-measure load torque outside the system, is f(x, t). The reconstructed dynamic model provides a basis for subsequent controller design. At the same time, by clarifying the disturbance source and observable variables, it provides support for constructing an Extended State Observer (ESO).

[0059] Step 2, Design of the non-linear scaling function: To reduce the fluctuation of the control input u(t), a family of continuously differentiable non-linear scaling functions g(z) is designed to scale and reconstruct the tracking error e x where e x = x(t) - r(t), and the expression of g(z) is:

[0060]

[0061] where is the independent variable of the non-linear scaling function, E is the allowable range of rotational speed error fluctuation, and c is a parameter to be calibrated.

[0062] Figure 2 shows how the parameter c to be calibrated affects the change of the non-linear scaling function curve.

[0063] g(z) can effectively solve the problem of large fluctuation of the control input u(t) under steady-state conditions. g(z) effectively reduces the noise level in the rotational speed signal by simultaneously reducing the feedback control gain and reconstructing the rotational speed signal fed back to the Extended State Observer. When |z| < 1, that is, when the tracking error |e x | < E, the value range of g(z) is between 0 and 1, and the closer the tracking error e x is to 0, the smaller the value of g(z); when |z| = 1, the derivative of g(z) exists and is 0, which provides the possibility for smooth switching between steady-state and transient modes. Figure 2 shows the function curve of g(z) and the change of its first derivative.

[0064] By introducing the non-linear scaling function g(z), the present invention reduces the gain of the proportional controller when the tracking error e x is small, thus significantly reducing the fluctuation of the steady-state control input u(t).

[0065] Step 3, Design of the order-elevating and order-reducing Extended State Observer:

[0066] The application provides a lifting step extended state observer for real-time estimation of system total disturbance f(x, t) and its first derivative. The observer combines a nonlinear scaling function, effectively reducing the influence of noise on observation accuracy, while improving the dynamic response capability of the system.

[0067] State variable reconstruction: introduce a nonlinear scaling function g(z) to scale the system tracking error e x to reduce noise level. The reconstructed error e xm and the reconstructed state variable x m are represented as:

[0068] e xm = g(z)e x

[0069] x m = g(z)e x + r(t)

[0070] where e x is the system tracking error, r is the target speed reference signal; g(z) is a nonlinear scaling function, x m is the reconstructed state variable. When the error e x is within the allowable fluctuation range (i.e., |e x | < E), the scaling function g(z) effectively reduces the high-frequency noise in the tracking error; when |e x | ≥ E, g(z) = 1, and the state variable is not affected by scaling.

[0071] Construction of the lifting step extended state observer (ARESO) based on the nonlinear scaling function:

[0072] In the lifting step extended state observer, the total disturbance f and its derivative f d are estimated as state variables. The dynamic equation of the observer is defined as follows:

[0073]

[0074] and are the estimated values of the total disturbance f(x, t) and its first derivative f d . β1 = ω0 2 , β2 = 2ω o are observer bandwidth related parameters. ω o represents the bandwidth of the lifting step extended state observer.

[0075] Elimination of state variable derivative:

[0076] To avoid directly using the state variable derivative (such as ) caused by high-frequency noise problems, the present application eliminates this effect by introducing intermediate variables ξ1 and ξ2.

[0077]

[0078] And further through mathematical transformation to get the following final form of the rising and falling order extended state observer:

[0079]

[0080] Through the design of the above rising and falling order extended state observer, the present application not only effectively reduces the influence of noise on disturbance estimation, but also significantly improves the disturbance response capability of the system. This improvement is largely due to the introduction of the nonlinear scaling function g(z), which plays a key role in the entire observer design.

[0081] Specifically, the nonlinear scaling function g(z) dynamically scales the tracking error e x , so that when the error is small, the amplitude of the scaled signal is significantly reduced, thereby reducing the interference of high-frequency noise on the estimation result. This scaling mechanism ensures that the extended state observer can ensure estimation accuracy while avoiding noise amplification problems. In addition, the continuous differentiability of the scaling function and the design of the derivative being continuous and zero at the transient and steady state operating point switching point provide a smooth transition when the error approaches the boundary of the allowed fluctuation range, further improving the stability of the system.

[0082] Through nonlinear reconstruction of state variables and error signals, the nonlinear scaling function g(z) limits high-frequency noise to the input end of the observer, avoiding the transmission and amplification of noise inside the observer. Combined with the scaling characteristics of g(z), the observer can reduce the control gain in the small error range, significantly reducing the control input fluctuation while maintaining good disturbance suppression capability.

[0083] In addition, the introduction of the nonlinear scaling function also optimizes the performance of the observer in the dynamic response stage, so that the system can quickly recover to a stable state when facing external disturbances or load mutations. Its combination with the low-gain control strategy not only further reduces the oscillation amplitude of the control input, but also improves the overall robustness and efficiency of the control system.

[0084] In summary, the application of the nonlinear scaling function g(z) is the key to the design of the rising and falling order extended state observer of the present application, which plays an indispensable and important role in noise reduction, stability improvement and dynamic performance optimization. This innovative design provides a more efficient and robust solution for diesel engine speed control. The observer can be combined with a low-gain control strategy to further optimize control performance.

[0085] Step 4, low-gain control law design:

[0086] The application proposes a low-gain control law, aiming to significantly reduce the fluctuation amplitude of control input while ensuring disturbance suppression capability, thereby improving the stability and economy of diesel engine speed control.

[0087] Based on the diesel engine dynamics model, the low-gain control law (LGC) designed by the application can be expressed as:

[0088]

[0089] Wherein, u(t) is the control input, u(t) is fed back to the diesel engine speed dynamic model of step 1, forming a closed-loop system from the reference speed to the actual diesel engine speed, k p = ω c , k p is the gain of the proportional controller in the low-gain control law, ω c is the pole position of the closed-loop system after pole placement, which will directly affect the stability of the closed-loop system, The total disturbance estimated by the extended state observer of the lifting order.

[0090] The low-gain control law effectively reduces the control gain when the error is small by multiplying the controller gain k p with the nonlinear scaling function g(z), thereby reducing the fluctuation of the steady-state control input.

[0091] In the design part of the low-gain control law, the introduction of the nonlinear scaling function enables the control gain to be dynamically adjusted within the error range, effectively reducing the oscillation amplitude of the control input during steady-state operation, reducing energy consumption and noise; when the external load suddenly changes or the disturbance is large, the control law can quickly increase the gain to ensure that the system has high disturbance suppression capability and achieves fast and stable dynamic response; the combination of the low-gain control strategy and the extended state observer of the lifting order further enhances the system's adaptability to nonlinear characteristics and uncertain disturbances.

[0092] The above only describes the preferred embodiments of the application, and it should be noted that for those skilled in the art, without departing from the principles of the application, several improvements and refinements can be made, which should also be considered within the scope of protection of the application.

Claims

1. A diesel engine speed active disturbance rejection control method based on nonlinear scaling, characterized in that, Includes the following steps: Step 1, Construct a dynamic model of diesel engine speed: The controlled output of the dynamic model of diesel engine speed is , , The diesel engine speed is the controlled input. , , The diesel engine indicated torque is represented by a dynamic model of the diesel engine speed, which includes a total disturbance. ; The dynamic model of diesel engine speed in step 1 is as follows: in: For controlled output, , This refers to the diesel engine speed; To control the input, To control the gain, , It is the moment of inertia; , , and There are three coefficients; For total disturbance; Step 2, construct the non-linear scaling function To address tracking errors Perform scaling and reconstruction. , The reconstructed state variables are based on the reference diesel engine speed. ; The nonlinear scaling function in step 2 is: in, , To allow for the range of speed error fluctuations, These are the parameters to be calibrated; Step 3, based on the reconstructed variables from Step 2 Construct an ascending-descending extended state observer to output the total perturbation. The estimated value The dynamic model of diesel engine speed fed back to step 1 is used as the total disturbance. ; In step 3, the ascending-descending expansion state observer is: in, , For observer bandwidth-related parameters, The total perturbation estimated by the ascending and descending order extended state observer. Total disturbance The estimated value of the first derivative, and As an intermediate variable; Step 4: Obtain the total perturbation based on the nonlinear scaling function from Step 2 and Step 3. The estimated value Construct a low-gain control law and output This feedback is then fed back to the dynamic model of diesel engine speed in step 1, forming a closed-loop system. In step 4, the low-gain control law is: Obtained from step 3, This represents the gain of the proportional controller in a low-gain control law.

2. The diesel engine speed active disturbance rejection control method based on nonlinear scaling as described in claim 1, characterized in that, , , This represents the bandwidth of the ascending and descending order extended state observer.

3. The diesel engine speed active disturbance rejection control method based on nonlinear scaling as described in claim 1, characterized in that, , It is the pole position of the closed-loop system after pole configuration.

Images