Nonlinear sliding mode predictive control method, device, equipment and storage medium

By constructing object model groups and calculating state estimation values ​​using extended Kalman filters, combining sliding mode control and prediction control, the robustness problem of nonlinear sliding mode prediction control method under unmeasurable disturbance is solved, and nonlinearity is achieved Improved system stability and robustness.

CN115598973BActive Publication Date: 2024-12-06CHINA PETROLEUM & CHEMICAL CORP +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110718004.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-06-28
Publication Date
2024-12-06
Estimated Expiration
2041-06-28

AI Technical Summary

Technical Problem

The existing nonlinear sliding mode prediction control methods are poorly robust under the influence of unmeasurable disturbances, making it difficult to effectively control nonlinear systems.

Method used

By constructing a group of object models for nonlinear systems, using an extended Kalman filter to calculate state estimates, generate switching functions, and bring them into the sliding mode control algorithm to optimize control inputs, combine sliding mode control and predictive control, and add switching functions. Constraints to improve system stability.

Benefits of technology

It effectively reduces the impact of unfathomable perturbations on nonlinear system prediction control and improves the robustness and stability of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115598973B_ABST
    Figure CN115598973B_ABST
Patent Text Reader

Abstract

The present invention discloses a nonlinear sliding mode predictive control method, device, equipment and readable storage medium, wherein the method comprises: constructing an object model group of a nonlinear system through model identification; generating a dynamic equation of the object model group, and calculating the disturbance characteristics of the object model group according to the dynamic equation; respectively obtaining the actual output measurement value and state estimation value of the nonlinear system at the current moment; using an extended Kalman filter, recursively calculating the optimal state estimation value of the nonlinear system at the next moment according to the actual output measurement value and the state estimation value; obtaining a corresponding switching function according to the optimal state estimation value; bringing the switching function into the dynamic equation of the object model group and obtaining the optimal solution according to the sliding mode control algorithm; the optimal solution is used as the input of the object model group. The present invention effectively reduces the influence of unmeasurable disturbances on the nonlinear predictive control system and improves the robustness of the nonlinear system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of industrial control, and in particular to a nonlinear sliding mode predictive control method, device, equipment and storage medium. Background Art

[0002] Most traditional industrial controls use a hierarchical structure to solve economic optimization and control problems. In the hierarchical structure, the first layer is the real-time optimization layer; the second layer is the supervision layer, which commonly uses nonlinear sliding mode predictive control; and the third layer is the conventional control layer, which includes most single-input single-output control loops.

[0003] The real-time optimization layer takes the maximum operating profit or the minimum operating cost as the control target, obtains the optimal solution through optimization calculation, and the nonlinear sliding mode predictive controller uses this as the set value. In each of its own operating cycles, the nonlinear sliding mode predictive controller will perform an optimization calculation based on the information transmitted by the underlying PID (Proportion Integration Differentiation) loop to obtain the control action to be transmitted to the underlying PID controller in the next cycle. After that, the underlying PID controller uses the control action as the set value of the underlying control loop, performs corresponding control on the underlying controlled object, and obtains the actual output value.

[0004] Chemical processes are generally nonlinear. For systems with weaker nonlinearity, the use of linear prediction models can fully approximate their dynamic processes, and linear prediction control has a better use effect; for strong nonlinear systems, linear models are often difficult to describe their complex dynamic processes, which makes linear prediction control difficult to apply. Traditional nonlinear sliding mode predictive control methods have problems such as model mismatch and unmeasurable disturbances. Model mismatch refers to the difference between the prediction model and the actual object. There are two main reasons for model mismatch: on the one hand, due to the influence of modeling errors, there is a large difference between the nonlinear prediction model and the actual object; on the other hand, changes in the external environment will cause changes in model parameters.

[0005] After research, the inventor found that the prior art has at least the following defects:

[0006] In practical applications, the nonlinear sliding mode predictive control method in the prior art has poor robustness, including: when the model of the nonlinear system is disturbed by unmeasurable disturbances, its predictive control effect will be greatly affected. Summary of the invention

[0007] The main purpose of the present invention is to reduce the influence of unmeasurable disturbance on the predictive control effect of nonlinear system.

[0008] To achieve the above object, the technical solution adopted by the present invention is:

[0009] The present invention discloses a nonlinear sliding mode predictive control method, comprising:

[0010] S11, constructing an object model group of a nonlinear system by model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model;

[0011] S12, generating a dynamic equation of the object model group, and calculating a disturbance characteristic of the object model group according to the dynamic equation;

[0012] The actual output measurement value and the state estimation value of the nonlinear system at the current moment are respectively obtained; the state estimation value obtaining method comprises: taking the model parameters of the object model group as variables at the moment before the current moment, and calculating and generating the state estimation value for estimating the state of the nonlinear system at the current moment according to the dynamic equation and its disturbance characteristics;

[0013] S14, using an extended Kalman filter, recursively calculating the optimal state estimate of the nonlinear system at the next moment according to the actual output measurement value and the state estimate;

[0014] S15, obtaining a corresponding switching function according to the optimal state estimation value;

[0015] S16, bringing the switching function into the dynamic equation of the object model group and obtaining an optimal solution according to a sliding mode control algorithm; the optimal solution is used as an input of the object model group.

[0016] Preferably, in the present invention, the optimal state estimation value of the nonlinear system at the next moment is calculated recursively according to the actual output measurement value and the state estimation value by using an extended Kalman filter, including:

[0017] S21, obtaining the optimal state estimate of the nonlinear system at time k-1

[0018] S22, obtaining the actual output measurement value {y of the nonlinear system at time k by detection k}, and through Obtain the state estimate of the nonlinear system at time k

[0019] S23, through the actual output measurement value {y k}Estimated value of the state Correction is performed to obtain the optimal state estimate of the nonlinear system at time k

[0020] Preferably, in the present invention, the switching function includes: in, is the optimal state estimate; C is a constant matrix.

[0021] Preferably, in the present invention, the step of bringing the switching function into the dynamic equation of the object model group and obtaining the optimal solution according to the sliding mode control algorithm includes:

[0022] The dynamic equations of the object model group include:

[0023] X k =f[X k-1 ,u k-1 , k-1, p k-1 ]+E[w k-1 ]

[0024] Z k =h[X k ,u k , k]

[0025] The performance optimization index of the nonlinear system is:

[0026] The constraints of the dynamic equations include:

[0027] X k ∈X,i=0,1,...,N-1

[0028] u k ∈U,i=0,1,...,N-1

[0029]

[0030] ρ∈[0,1)

[0031] Among them, ρ∈[0,1) is the shrinkage rate parameter; the weighting matrix is a positive symmetric matrix; the state constraint set X and the input constraint set U; a k is the switching function; a (0|k) is the zeroth switching function at time k, a (N|k) is the Nth switching function at time k, and u(·) is the control sequence.

[0032] Preferably, in the present invention, the step of bringing the switching function into the dynamic equation of the object model group and obtaining the optimal solution according to the sliding mode control algorithm comprises the steps of:

[0033] S31, given the initial state x(0), N is the prediction time domain, the shrinkage rate ρ∈[0,1), the state constraint set X and the input constraint set U; in addition,

[0034] S32, set k=0;

[0035] S33, the optimal state estimation value at the current time k according to the dynamic equation of the object model group Get the corresponding switching function a k ;

[0036] S34, the switching function a at the current k moment k Substitute into the dynamic equation of the object model group; obtain the optimal control sequence u according to the sliding mode predictive control algorithm through the performance optimization index and the constraint conditions * (·)={u * (0|k),...,u * (N-1|k)}; Determine the control action u from the optimal control sequence * (0|k) is the optimal solution at the current time k;

[0037] S35, when k is less than the preset value, set k=k+h, and return to step S33; h is the number of interval time, and 1≤h≤N;

[0038] S36: When k reaches the preset value, the calculation for obtaining the optimal solution is terminated.

[0039] Preferably, in the present invention, the controlled object sub-object model includes:

[0040] a first plant sub-plant model and a second plant sub-plant model.

[0041] Preferably, in the present invention, the calculating the disturbance characteristics of the object model group includes:

[0042] The state space model of the object model group is subjected to first-order Taylor expansion at the nominal model parameter point multiple times according to the preset time interval with the model parameters as variables, and the statistical characteristics of the disturbance are calculated.

[0043] Preferably, in the present invention, performing first-order Taylor expansion on the state space model of the object model group with the model parameters as variables at the nominal model parameter points multiple times according to preset time intervals to calculate the statistical characteristics of the disturbance includes:

[0044] Assuming that the parameter variables of the object model group obey normal distribution, the changes of the parameters of the object model group are equivalent to disturbances, and the statistical characteristics of the disturbances are calculated.

[0045] Preferably, in the present invention, the model parameters include:

[0046] The liquid level value, flow value, temperature value or pressure value of the nonlinear system.

[0047] In another aspect of the present invention, a nonlinear sliding mode predictive control device is provided, comprising:

[0048] A grouping unit, used for constructing an object model group of a nonlinear system through model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model;

[0049] a characteristic calculation unit, configured to generate a dynamic equation of the object model group, and calculate a disturbance characteristic of the object model group according to the dynamic equation;

[0050] A numerical acquisition unit is used to acquire the actual output measurement value and state estimation value of the nonlinear system at the current moment; the acquisition of the state estimation value comprises: taking the input acting on the object model group as a parameter, calculating the state estimation value of the current moment generated at the previous moment according to the dynamic equation and its disturbance characteristics;

[0051] A correction unit, configured to calculate the optimal state estimation value of the nonlinear system at the next moment in a recursive manner according to the actual output measurement value and the state estimation value through an extended Kalman filter;

[0052] A switching function generating unit, used for obtaining a corresponding switching function according to the optimal state estimation value;

[0053] The result generating unit is used to bring the switching function into the dynamic equation of the object model group and obtain the optimal solution according to the sliding mode control algorithm; the optimal solution is used as the input of the object model group.

[0054] In another aspect of the embodiment of the present invention, a nonlinear sliding mode predictive control device is also provided, including:

[0055] Memory for storing computer programs;

[0056] A processor is used to call and execute the computer program to implement each step of the nonlinear sliding mode predictive control method as described in any one of the above items.

[0057] In another aspect of the embodiment of the present invention, a storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, each step of the nonlinear sliding mode predictive control method as described in any one of the above items is implemented.

[0058] Beneficial Effects

[0059] The present invention combines sliding mode control with predictive control, and uses a switching function as a new variable to substitute into the sliding mode predictive control for optimization and solution; wherein, the switching function and the control input are used as optimization variables in the objective function, and switching function constraints are added, and the control effect as the optimal solution can be obtained by minimizing the designed objective function; the present invention uses the switching function in the sliding mode control as a new variable, and punishes the deviation of the switching function and the control input relative to the equivalent control in the objective function, so that the objective function is zero, which means that the system state reaches the sliding mode surface, and the control law is equal to the equivalent control; since the present invention combines the advantages of both sliding mode control and predictive control methods, a sliding mode predictive controller with stability guarantee is designed, which can also have stability when the nonlinear system has large unmeasurable disturbances, thereby effectively reducing the influence of unmeasurable disturbances on the predictive control effect of the nonlinear system and improving the robustness of the nonlinear system.

[0060] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application and implement it according to the contents of the specification, and at the same time to make the above and other purposes, technical features and advantages of the present application easier to understand, one or more preferred embodiments are listed below and described in detail with the accompanying drawings as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0062] Figure 1 It is a schematic diagram of the steps of the nonlinear sliding mode predictive control method described in the present invention;

[0063] Figure 2 It is a structural schematic diagram of the nonlinear system described in the present invention;

[0064] Figure 3 Schematic diagram of the distribution of model parameters in the present invention;

[0065] Figure 4 It is a structural schematic diagram of the nonlinear sliding mode predictive control device of the present invention;

[0066] Figure 5 It is a schematic diagram of the structure of the nonlinear sliding mode predictive control device described in the present invention. DETAILED DESCRIPTION

[0067] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0068] Embodiment 1

[0069] In order to effectively reduce the impact of unmeasurable disturbances on the predictive control effect of nonlinear systems, refer to Figure 1 and Figure 2 , an embodiment of the present invention provides a nonlinear sliding mode predictive control method, comprising:

[0070] S11, constructing an object model group of a nonlinear system by model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model;

[0071] In the embodiment of the present invention, the object model group of the nonlinear system is first defined, and the object models of PID 21, actuator 22 and controlled object 23 in the nonlinear system are used as sub-object models of the object model group. The nonlinear sliding mode predictive controller 01 can generate input variables u of the object model group.

[0072] It should be noted that, in the embodiment of the present invention, according to different application scenarios and requirements of the nonlinear sliding mode predictive control, the controlled object 23 in the nonlinear system may include one or more objects. Therefore, correspondingly, the controlled object 23 in the embodiment of the present invention may also include one or more objects; for example, Figure 2 As shown, when the controlled object 23 includes a first controlled object and a second controlled object, the controlled object sub-object model may include the first controlled object sub-object model and the second controlled object sub-object model.

[0073] S12, generating a dynamic equation of the object model group, and calculating a disturbance characteristic of the object model group according to the dynamic equation;

[0074] Then, the object model group (including the PID sub-object model, the actuator sub-object model and the controlled object sub-object model) is taken as a whole to generate the corresponding dynamic equation, and the disturbance characteristics of the object model group are calculated according to the dynamic equation.

[0075] In practical applications, the method for calculating the disturbance characteristics of the object model group can be specifically to perform a first-order Taylor expansion on the state space model of the object model group at the nominal model parameter point with the model parameters as variables at preset time intervals for multiple times, thereby calculating the statistical characteristics of the disturbance. Preferably, it can be assumed that the parameter variables of the object model group obey the normal distribution, the changes in the parameters of the object model group are equivalent to the disturbance, and the statistical characteristics of the disturbance are calculated. In practical applications, the model parameters can be the liquid level value, flow value, temperature value or pressure value measured or collected in a nonlinear system.

[0076] S13, respectively obtaining the actual output measurement value and state estimation value of the nonlinear system at the current moment; the method for obtaining the state estimation value comprises: taking the model parameters of the object model group as variables at a moment before the current moment, and calculating and generating the state estimation value for estimating the state of the nonlinear system at the current moment according to the dynamic equation and its disturbance characteristics.

[0077] In the embodiment of the present invention, not only the state estimation value of the nonlinear system is calculated, but also the actual output measurement value at the same time is collected. Specifically, for time k, on the one hand, the state estimation value of the nonlinear system at time k can be generated at time k-1, and on the other hand, the actual output measurement value y of the controlled object at time k needs to be collected; wherein, the acquisition method of the state estimation value at time k may include: at time k-1 (i.e., the moment before the current moment), the input u acting on the object model group is used as a parameter, and the state estimation value for estimating the nonlinear system at time k is calculated according to the dynamic equation and its disturbance characteristics. In practical applications, when the application scenario of the nonlinear sliding mode predictive control system is to control the absorption tower liquid level system, the controlled object may specifically be the absorption tower liquid level; when the application scenario of the nonlinear sliding mode predictive control system is to control the heating furnace outlet temperature system, the controlled object may specifically be the heating furnace outlet temperature.

[0078] S14, using an extended Kalman filter, recursively calculating the optimal state estimate of the nonlinear system at the next moment according to the actual output measurement value and the state estimate;

[0079] In the embodiment of the present invention, an extended Kalman filter 03 is further provided to obtain the state estimation value of the object model group, and the state estimation value is corrected by the actual output measurement value to obtain the optimal state estimation value, thereby effectively avoiding the error between the state estimation value and the actual state of the nonlinear system being too large; in practical applications, the optimal state estimation value of the nonlinear system at the next moment is calculated in a recursive manner, which may specifically include the following steps:

[0080] S21, obtaining the optimal state estimate of the nonlinear system at time k-1

[0081] S22, obtaining the actual output measurement value {y of the nonlinear system at time k by detection k}, and through Obtain the state estimate of the nonlinear system at time k

[0082] S23, through the actual output measurement value {y k}Estimated value of the state Correction is performed to obtain the optimal state estimate of the nonlinear system at time k

[0083] At time k, the output u of the nonlinear predictive control is applied to the generalized object model to obtain the actual output value y of the system; then, the predicted state in the nonlinear predictive control is and the actual output value y of the system at the current moment are substituted into the Kalman filter to obtain the optimal estimated state of the system at this time Finally, this state is substituted into the nonlinear predictive control algorithm for optimization and solution to obtain the control action at the next moment.

[0084] The specific implementation steps of the above algorithm are described below.

[0085] 1. Compute the state space model of generalized objects

[0086] Assume that after model identification, the generalized object model of the discrete nonlinear system is obtained:

[0087] X k =f[X k-1 ,u k-1 , k-1, p k-1 ] (1)

[0088] Z k =h[X k ,u k , k]

[0089] Where f[·] is an n-dimensional vector nonlinear function, and h[·] is an m-dimensional vector nonlinear function. k is the system status, u k is the system input, p k is the model parameter, Z k Output for the system.

[0090] 2. Treat the changes in model parameters as disturbances

[0091] Model mismatch and unmeasurable disturbances usually exist in nonlinear process control. In order to analyze the impact of model mismatch on control effect, here we only analyze the situation when the system model parameters change or there are unmeasurable disturbances.

[0092] Assumptions Figure 2 The model parameter variables follow the normal distribution. The model parameter variables are represented by vector P, and their mean and variance are as follows:

[0093] E[P]=P0

[0094] E[(P-P0)(P-P0) T ]=Q P

[0095] Where P0 represents the nominal model parameter value, Q P Represents the covariance matrix of the model parameters, assuming that the actual parameters after the change are P i,plant , the model parameters are distributed as Figure 3 shown.

[0096] Figure 3 The dashed envelope represents the nominal model parameter P i,0 The probability distribution curve of the actual model parameter P i,plant With the nominal model parameter P i,0 The relationship can be expressed as P i,plant =P i,0 +δ i , that is, the change is δ i .

[0097] In order to estimate the state of the system after the model parameters change, the change of the model parameters can be equivalent to the disturbance, and the nonlinear prediction model is expanded by the first order Taylor with the model parameters as variables at the nominal model parameter point. In order to make the output of the prediction model close to the output of the actual system, it is necessary to calculate the statistical characteristics of the disturbance based on the uncertainty information of the model parameters.

[0098] In the nominal model parameter P i,0 Performing a first-order Taylor expansion on the discrete state space model (1), the dynamic system equation is obtained as follows:

[0099]

[0100]

[0101] The above formula can be simplified to:

[0102] X k =f[X k-1 ,u k-1 , k-1, p k-1 ]+w k-1

[0103] Z k =h[X k ,u k , k]+vk

[0104] Among them, w k-1 、v k is a disturbance equivalent to the uncertainty in the model parameters.

[0105]

[0106]

[0107] 3. Design of extended Kalman filter after equating system disturbance to Gaussian white noise

[0108] (1) Find w k-1 、v k Statistical properties of

[0109] w k-1 The mean value of k-1 ], and E[w k-1 ]≠0, w k-1 The variance of is:

[0110] E[(w k-1 -E[w k-1 ])·(w j-1 -E[w j-1 ]) T ]=Q k-1

[0111] v k The mean value is E[v k ], and E[v k ]=0,v k The variance of is:

[0112] E[(v k -E[v k ])·(v j -E[v j ]) T ]=R k

[0113] and

[0114] The above formula can be simplified to:

[0115] w k-1 ~(Ew k-1 ], Q k-1 )

[0116] v k ~(0,0)

[0117] (2) w k-1 Equivalent to Gaussian white noise with zero mean

[0118] make is the equivalent zero-mean Gaussian white noise

[0119] The original nonlinear system equation can be equivalent to:

[0120]

[0121] Z k =h[X k ,u k , k]+v k

[0122] in, The mean and variance of are as follows:

[0123]

[0124]

[0125] v k The mean and variance of are as follows:

[0126] E[v k ]=0,

[0127] E[(v k -E[v k ])·(v j -E[v j ]) T ]=R k =0

[0128] and

[0129] (3) Design a Kalman filter for state estimation

[0130] The nonlinear function f[X k-1 ,u k-1 , k-1, p k-1 ] and h[X k ,u k , k] are approximately and The first-order Taylor polynomial nearby is:

[0131]

[0132]

[0133] F k-1 and H k is the Jacobian matrix:

[0134]

[0135]

[0136] This gives a linear state-space model approximated by first-order Taylor polynomials:

[0137]

[0138]

[0139] Using the Extended Kalman Filter (EKF) algorithm:

[0140] initialization:

[0141]

[0142]

[0143] Prediction status:

[0144]

[0145] State prediction error covariance matrix:

[0146]

[0147] in,

[0148]

[0149] Kalman gain:

[0150]

[0151]

[0152]

[0153] State Estimation:

[0154]

[0155] State estimation error covariance matrix:

[0156] P k =(IK k H k ) k|k-1

[0157] After solving, we can get the optimal state estimate

[0158] S15, obtaining a corresponding switching function according to the optimal state estimation value;

[0159] The inventive idea of ​​the embodiment of the present invention includes: utilizing the respective advantages of sliding mode control and predictive control methods to design a sliding mode predictive controller with stability guarantee. The switching function in the sliding mode control is selected as a new variable, and the deviation of the switching function and the control input relative to the equivalent control is penalized in the objective function, so that the objective function is zero, which means that the system state reaches the sliding surface and the control law is equal to the equivalent control. By minimizing the designed objective function, the numerical solution of the optimal control action can be obtained.

[0160] For the predictive control method of nonlinear system, stability guarantee is a prerequisite for its application in practice. If infinite time domain performance index is used, there will be infinite optimization variables in the optimization problem and it cannot be solved. Therefore, in the embodiment of the present invention, the optimization problem of finite time domain performance index is used.

[0161] In the sliding control algorithm, it is first necessary to select appropriate switching function parameters so that the system has asymptotically stable sliding mode dynamics on the sliding surface, and then obtain the explicit state feedback control law according to the arrival condition. By implementing the obtained state feedback control law, the system state will reach the sliding surface in a finite time and start the sliding mode motion. Since the arrival condition is met, when the system state reaches the sliding surface, it will no longer leave the sliding surface, but will asymptotically stabilize to the equilibrium point along the sliding surface.

[0162] In the embodiment of the present invention, the switching function (linear switching function) may be: in, is the optimal state estimate; C is a constant matrix, then the corresponding sliding surface is a=Cx=0. Define the following null space: A:N(C)={x|||a||=||Cx||≤Δ, C∈constant matrix},

[0163] Among them, the positive constant Δ determines the bandwidth of the quasi-sliding mode.

[0164] For discrete-time nonlinear systems, the design steps of the sliding mode controller can be as follows:

[0165] First, select the switching function a k The corresponding sliding mode has asymptotically stable characteristics; the switching function a in the embodiment of the present invention k It can be chosen as a linear form with respect to state x, that is For the linear sliding surface under such a linear switching function, its parameter C is very easy to obtain. In addition, the switching function in the embodiment of the present invention can also be selected as a nonlinear form, and those skilled in the art can set it according to the technology in the field, and no further explanation is given here.

[0166] S16, bringing the switching function into the dynamic equation of the object model group and obtaining an optimal solution according to a sliding mode control algorithm; the optimal solution is used as an input of the object model group.

[0167] Specifically, the switching function a at the current time k can be k Substitute into the dynamic equation of the object model group; then use the sliding mode predictive control algorithm through the performance optimization index and the constraints to obtain the corresponding control action (i.e., state feedback control action) according to the arrival condition. In practical applications, the arrival condition can be: ||a(k+1)||<||a(k)||.

[0168] The optimization problem of sliding mode predictive control of nonlinear systems based on shrinkage constraints can be specifically described as follows:

[0169] Assume the dynamic equation of the object model group is:

[0170] X k =f[X k-1 ,u k-1 , k-1, p k-1 ]+E[w k-1 ]

[0171] Z k =h[X k ,u k , k]

[0172] The performance optimization index of the nonlinear system is

[0173] In the embodiment of the present invention, a special switching function constraint is added to the optimization problem of the traditional sliding mode predictive control method to make the closed-loop system stable.

[0174] The constraints include:

[0175] X k ∈X,i=0,1,...,N-1

[0176] u k ∈U,i=0,1,...,N-1

[0177]

[0178] ρ∈[0,1)

[0179] Wherein, ρ∈[0,1) is the shrinkage rate parameter, which is a fixed value set by technicians in this field according to actual needs and / or experience; the weighting matrix is a positive symmetric matrix; the state constraint set X and the input constraint set U; a k is the switching function; a (0|k)is the zeroth switching function at time k, a (N|k) is the Nth switching function at time k, and u(·) is the control sequence. The shrinkage rate parameter ρ is added as a new optimization variable to the optimization problem, which can obtain better control performance and increase the feasible domain of the algorithm.

[0180] Preferably, in order to obtain better control performance and increase the feasible domain of the algorithm, in the embodiment of the present invention, a step of determining the optimal shrinkage rate parameter ρ value may also be included. Specifically, when setting the value of the shrinkage rate parameter ρ, the ρ value is adjusted according to the deviation between the prediction result of the previous moment (or the previous few moments) of the prediction system and the actual result; in practical applications, when the deviation between the prediction result and the actual result is too large, the value of ρ needs to be increased, even if its value is closer to 1, so as to improve the subsequent prediction accuracy; and when the deviation between the prediction result and the actual result is too small, the value of ρ needs to be reduced, even if its value is closer to 0, so as to reduce unnecessary calculation amount; that is, by determining the optimal shrinkage rate parameter, a balance can be achieved between accuracy and calculation amount, and then the real-time performance of the prediction can be improved by reducing the calculation cost under the premise of ensuring a certain prediction accuracy. It should be noted that those skilled in the art can set a reasonable threshold range of the deviation degree according to actual needs, and determine whether the deviation is too large or too small by comparing it with the deviation degree between the prediction result and the actual result.

[0181] In the embodiment of the present invention, the specific steps of obtaining the optimal solution according to the sliding mode control algorithm may be as follows:

[0182] S31, given the initial state x(0), N is the prediction time domain, the shrinkage rate ρ∈[0,1), the state constraint set X and the input constraint set U; in addition,

[0183] S32, set k=0;

[0184] S33, the optimal state estimation value at the current time k according to the dynamic equation of the object model group Get the corresponding switching function a k ;

[0185] S34, the switching function a at the current k moment k Substitute into the dynamic equation of the object model group; obtain the optimal control sequence u according to the sliding mode predictive control algorithm through the performance optimization index and the constraint conditions * (·)={u * (0|k),...,u * (N-1|k)}; Determine the control action u from the optimal control sequence * (0|k) is the optimal solution at the current time k;

[0186] S35. When k is less than the preset value, set k=k+h, and return to step S33; wherein h is the number of interval moments, and 1≤h≤N; it should be noted that the value of h can be set by technical personnel in this field according to actual needs. The larger the h value is, the more the number of moments between each calculation process is, and therefore, the smaller the overall calculation amount is; the smaller the h value is, the fewer the number of moments between each calculation process is, and the higher the accuracy of the nonlinear sliding mode prediction system in the embodiment of the present invention is.

[0187] S36: When k reaches the preset value, the calculation for obtaining the optimal solution is terminated.

[0188] In the present invention, the optimal state estimation value is obtained in the following manner: taking a nonlinear system at a certain moment (k moment) as an example, on the one hand, the state estimation value of the nonlinear system at k moment is obtained at the previous moment (k-1) of the moment; on the other hand, the actual output measurement value y of the controlled object at k moment is also obtained; then, the state estimation value is corrected according to the actual output measurement value by an extended Kalman filter to generate the optimal state estimation value of the nonlinear system. In practical applications, when the application scenario of the nonlinear sliding mode predictive control system is to control the absorption tower liquid level system, the controlled object can specifically be the absorption tower liquid level; when the application scenario of the nonlinear sliding mode predictive control system is to control the heating furnace outlet temperature system, the controlled object can specifically be the heating furnace outlet temperature.

[0189] To summarize, in an embodiment of the present invention, the PID sub-object model, the actuator sub-object model and the controlled object sub-object model are first defined as an object model group, and then the disturbance characteristics of the object model group are calculated and obtained; then, the extended Kalman filter algorithm is used to obtain the optimal state estimate of the object model group. Next, the embodiment of the present invention combines the sliding mode control with the predictive control, and uses the switching function as a new variable to substitute into the sliding mode predictive control for optimization and solution; wherein, the switching function and the control input are used as optimization variables in the objective function, and the switching function constraint is added, and the control effect as the optimal solution can be obtained by minimizing the designed objective function; the embodiment of the present invention uses the switching function in the sliding mode control as a new variable, and penalizes the deviation of the switching function and the control input relative to the equivalent control in the objective function, so that the objective function is zero, which means that the system state reaches the sliding mode surface, and the control law is equal to the equivalent control; since the embodiment of the present invention has both the advantages of the sliding mode control and the predictive control method, a sliding mode predictive controller with stability guarantee is designed, which can be stable even when the nonlinear system has large unmeasurable disturbances, thereby effectively reducing the influence of unmeasurable disturbances on the predictive control effect of the nonlinear system and improving the robustness of the nonlinear system.

[0190] Embodiment 2

[0191] In another aspect of the embodiment of the present invention, a nonlinear sliding mode predictive control device is also provided. Figure 4 A schematic diagram showing the structure of a nonlinear sliding mode predictive control device provided by an embodiment of the present invention is shown. The nonlinear sliding mode predictive control device is Figure 1 The device corresponding to the nonlinear sliding mode predictive control method described in the corresponding embodiment, that is, the virtual device is used to implement Figure 1 In the nonlinear sliding mode predictive control method in the corresponding embodiment, each virtual module constituting the nonlinear sliding mode predictive control device can be executed by an electronic device, such as a network device, a terminal device, or a server. Specifically, the nonlinear sliding mode predictive control device in the embodiment of the present invention includes:

[0192] The group building unit 101 is used to build an object model group of a nonlinear system through model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model;

[0193] The characteristic calculation unit 102 is used to generate a dynamic equation of the object model group, and calculate the disturbance characteristic of the object model group according to the dynamic equation;

[0194] The numerical acquisition unit 103 is used to acquire the actual output measurement value and state estimation value of the nonlinear system at the current moment; the acquisition of the state estimation value includes: taking the input acting on the object model group as a parameter, calculating the state estimation value of the current moment generated at the previous moment according to the dynamic equation and its disturbance characteristics;

[0195] The correction unit 104 is used to calculate the optimal state estimation value of the nonlinear system at the next moment in a recursive manner according to the actual output measurement value and the state estimation value through an extended Kalman filter;

[0196] The switching function generating unit 105 is used to obtain a corresponding switching function according to the optimal state estimation value;

[0197] The result generating unit 106 is used to bring the switching function into the dynamic equation of the object model group and obtain the optimal solution according to the sliding mode control algorithm; the optimal solution is used as the input of the object model group.

[0198] Since the working principle and beneficial effects of the nonlinear sliding mode predictive control device in the embodiment of the present invention have been Figure 1 The corresponding nonlinear sliding mode predictive control method is also recorded and explained, so they can be cross-referenced and will not be repeated here.

[0199] Embodiment 3

[0200] Corresponding to the method embodiment, the present application also provides a nonlinear sliding mode predictive control device, such as a terminal, a server, etc. Among them, the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc., but is not limited thereto.

[0201] An example of a hardware structure block diagram of a nonlinear sliding mode predictive control device provided by an embodiment of the present invention is shown in FIG. Figure 5 As shown, it may include:

[0202] Processor 1, communication interface 2, memory 3 and communication bus 4;

[0203] The processor 1, the communication interface 2, and the memory 3 communicate with each other via the communication bus 4;

[0204] Optionally, the communication interface 2 may be an interface of a communication module, such as an interface of a GSM module; the processor 1 may be a central processing unit CPU, or an application specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement an embodiment of the present invention.

[0205] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0206] The processor 1 is specifically used to execute the computer program stored in the memory 3 to perform the following steps:

[0207] S11, constructing an object model group of a nonlinear system by model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model;

[0208] S12, generating a dynamic equation of the object model group, and calculating a disturbance characteristic of the object model group according to the dynamic equation;

[0209] S13, respectively obtaining the actual output measurement value and the state estimation value of the nonlinear system at the current moment; the method for obtaining the state estimation value comprises: taking the model parameters of the object model group as variables at the moment before the current moment, calculating and generating the state estimation value for estimating the state of the nonlinear system at the current moment according to the dynamic equation and its disturbance characteristics;

[0210] S14, using an extended Kalman filter, recursively calculating the optimal state estimate of the nonlinear system at the next moment according to the actual output measurement value and the state estimate value;

[0211] S15, obtaining a corresponding switching function according to the optimal state estimation value;

[0212] S16, bringing the switching function into the dynamic equation of the object model group and obtaining an optimal solution according to a sliding mode control algorithm; the optimal solution is used as an input of the object model group.

[0213] The nonlinear sliding mode predictive control device in the embodiment of the present invention, when the program instructions included in its computer program product are executed by a computer, can enable the computer to execute the nonlinear sliding mode predictive control methods described in the above aspects and achieve the same technical effects.

[0214] Embodiment 4

[0215] In an embodiment of the present invention, a storage medium is further provided, which may store a program suitable for execution by a processor, wherein the program is used to:

[0216] S11, constructing an object model group of a nonlinear system by model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model;

[0217] S12, generating a dynamic equation of the object model group, and calculating a disturbance characteristic of the object model group according to the dynamic equation;

[0218] S13, respectively obtaining the actual output measurement value and the state estimation value of the nonlinear system at the current moment; the method for obtaining the state estimation value comprises: taking the input acting on the object model group as a parameter at a moment before the current moment, and calculating and generating the state estimation value for estimating the state of the nonlinear system at the current moment according to the dynamic equation and its disturbance characteristics;

[0219] S14, using an extended Kalman filter, recursively calculating the optimal state estimate of the nonlinear system at the next moment according to the actual output measurement value and the state estimate value;

[0220] S15, obtaining a corresponding switching function according to the optimal state estimation value;

[0221] S16, bringing the switching function into the dynamic equation of the object model group and obtaining an optimal solution according to a sliding mode control algorithm; the optimal solution is used as an input of the object model group.

[0222] Optionally, the detailed functions and extended functions of the program may refer to the above description.

[0223] The above product can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the method provided by the embodiment of the present invention.

[0224] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0225] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0226] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0227] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0228] It should be understood that in the embodiments of the present application, the various embodiments and features can be combined with each other to solve the aforementioned technical problems.

[0229] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0230] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A nonlinear sliding mode predictive control method, characterized in that: include: S11, constructing an object model group of a nonlinear system by model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model; S12, generating a dynamic equation of the object model group, and calculating a disturbance characteristic of the object model group according to the dynamic equation; S13, respectively obtaining the actual output measurement value and state estimation value of the nonlinear system at the current moment; The method for acquiring the state estimation value comprises: taking the model parameters of the object model group as variables at a moment before the current moment, and calculating and generating the state estimation value for estimating the state of the nonlinear system at the current moment according to the dynamic equation and its disturbance characteristics; S14, using an extended Kalman filter, recursively calculating the optimal state estimate of the nonlinear system at the next moment according to the actual output measurement value and the state estimate value; S15, obtaining a corresponding switching function according to the optimal state estimation value; S16, bringing the switching function into the dynamic equation of the object model group and obtaining an optimal solution according to a sliding mode control algorithm; the optimal solution is used as an input of the object model group; The method of calculating the optimal state estimation value of the nonlinear system at the next moment in a recursive manner according to the actual output measurement value and the state estimation value by using an extended Kalman filter includes: S21, obtaining the optimal state estimate of the nonlinear system at time k-1 S22, obtaining the actual output measurement value {y k }, and through Obtain the state estimate of the nonlinear system at time k S23, through the actual output measurement value {y k }Estimated value of the state Correction is performed to obtain the optimal state estimate of the nonlinear system at time k The switching function includes: in, is the optimal state estimate; C is a constant matrix; Substituting the switching function into the dynamic equation of the object model group and obtaining the optimal solution according to the sliding mode control algorithm includes: The dynamic equations of the object model group include: X k =f[X k-1 ,u k-1 ,k-1,p k-1 ]+E[w k-1 ] Z k =h[X k ,u k ,k] The performance optimization index of the nonlinear system is: The constraints of the dynamic equations include: X k ∈X,i=0,1,...,N-1 u k ∈U,i=0,1,...,N-1 ρ∈[0,1) Among them, ρ∈[0,1) is the shrinkage rate parameter; the weighting matrix is a positive symmetric matrix; the state constraint set X and the input constraint set U; a k is the switching function; a (0|k) is the zeroth switching function at time k, a (N|k) is the Nth switching function at time k, and u(·) is the control sequence; The step of bringing the switching function into the dynamic equation of the object model group and obtaining the optimal solution according to the sliding mode control algorithm comprises the following steps: S31, given the initial state x(0), N is the prediction time domain, the shrinkage rate ρ∈[0,1), the state constraint set X and the input constraint set U; in addition, Q>0, R>0, S32, set k=0; S33, the optimal state estimation value at the current time k according to the dynamic equation of the object model group Get the corresponding switching function a k ; S34, the switching function a at the current k moment k Substitute into the dynamic equation of the object model group; obtain the optimal control sequence u according to the sliding mode predictive control algorithm through the performance optimization index and the constraint conditions * (·)={u * (0|k),...,u * (N-1|k)}; Determine the control action u from the optimal control sequence * (0|k) is the optimal solution at the current time k; S35, when k is less than the preset value, set k=k+h, and return to step S33; h is the number of interval time, and 1≤h≤N; S36: When k reaches the preset value, the calculation for obtaining the optimal solution is terminated.

2. The nonlinear sliding mode predictive control method according to claim 1, characterized in that: The controlled object sub-object model includes: a first plant sub-plant model and a second plant sub-plant model.

3. The nonlinear sliding mode predictive control method according to claim 1, characterized in that: The calculating the disturbance characteristics of the object model group includes: The state space model of the object model group is subjected to first-order Taylor expansion at the nominal model parameter point multiple times according to the preset time interval with the model parameters as variables, and the statistical characteristics of the disturbance are calculated.

4. The nonlinear sliding mode predictive control method according to claim 3, characterized in that: The method of performing first-order Taylor expansion on the state space model of the object model group at the nominal model parameter point multiple times according to the preset time interval with the model parameters as variables to calculate the statistical characteristics of the disturbance includes: Assuming that the parameter variables of the object model group obey normal distribution, the changes of the parameters of the object model group are equivalent to disturbances, and the statistical characteristics of the disturbances are calculated.

5. The nonlinear sliding mode predictive control method according to claim 3, characterized in that: The model parameters include: The liquid level value, flow value, temperature value or pressure value of the nonlinear system.

6. A nonlinear sliding mode predictive control device, characterized in that: include: A grouping unit, used for constructing an object model group of a nonlinear system through model identification; the object model group includes a PID sub-object model, an actuator sub-object model and a controlled object sub-object model; the object model group is expressed in the form of a state space model; a characteristic calculation unit, configured to generate a dynamic equation of the object model group, and calculate a disturbance characteristic of the object model group according to the dynamic equation; A value acquisition unit, used to respectively acquire an actual output measurement value and a state estimation value of the nonlinear system at a current moment; The method for obtaining the state estimation value comprises: using the model parameters of the object model group as variables at a moment before the current moment, calculating and generating the state estimation value for estimating the state of the nonlinear system at the current moment according to the dynamic equation and its disturbance characteristics; A correction unit, configured to calculate the optimal state estimation value of the nonlinear system at the next moment in a recursive manner according to the actual output measurement value and the state estimation value through an extended Kalman filter; A switching function generating unit, used for obtaining a corresponding switching function according to the optimal state estimation value; A result generating unit is used to bring the switching function into the dynamic equation of the object model group and obtain an optimal solution according to a sliding mode control algorithm; the optimal solution is used as an input of the object model group; The method of calculating the optimal state estimation value of the nonlinear system at the next moment in a recursive manner according to the actual output measurement value and the state estimation value by using an extended Kalman filter includes: S21, obtaining the optimal state estimate of the nonlinear system at time k-1 S22, obtaining the actual output measurement value {y k }, and through Obtain the state estimate of the nonlinear system at time k S23, through the actual output measurement value {y k }Estimated value of the state Correction is performed to obtain the optimal state estimate of the nonlinear system at time k The switching function includes: in, is the optimal state estimate; C is a constant matrix; Substituting the switching function into the dynamic equation of the object model group and obtaining the optimal solution according to the sliding mode control algorithm includes: The dynamic equations of the object model group include: X k =f[X k-1 ,u k-1 ,k-1,p k-1 ]+E[w k-1 ] Z k =h[X k ,u k ,k] The performance optimization index of the nonlinear system is: The constraints of the dynamic equations include: X k ∈X,i=0,1,...,N-1 u k ∈U,i=0,1,...,N-1 ρ∈[0,1) Among them, ρ∈[0,1) is the shrinkage rate parameter; the weighting matrix is a positive symmetric matrix; the state constraint set X and the input constraint set U; a k is the switching function; a (0|k) is the zeroth switching function at time k, a (N|k) is the Nth switching function at time k, and u(·) is the control sequence; The step of bringing the switching function into the dynamic equation of the object model group and obtaining the optimal solution according to the sliding mode control algorithm comprises the following steps: S31, given the initial state x(0), N is the prediction time domain, the shrinkage rate ρ∈[0,1), the state constraint set X and the input constraint set U; in addition, Q>0, R>0, S32, set k=0; S33, the optimal state estimation value at the current time k according to the dynamic equation of the object model group Get the corresponding switching function a k ; S34, the switching function a at the current k moment k Substitute into the dynamic equation of the object model group; obtain the optimal control sequence u according to the sliding mode predictive control algorithm through the performance optimization index and the constraint conditions * (·)={u * (0|k),...,u * (N-1|k)}; Determine the control action u from the optimal control sequence * (0|k) is the optimal solution at the current time k; S35, when k is less than the preset value, set k=k+h, and return to step S33; h is the number of interval time, and 1≤h≤N; S36: When k reaches the preset value, the calculation for obtaining the optimal solution is terminated.

7. A nonlinear sliding mode predictive control device, comprising: Memory for storing computer programs; A processor is used to call and execute the computer program to implement each step of the nonlinear sliding mode predictive control method as described in any one of claims 1 to 5.

8. A storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the nonlinear sliding mode predictive control method according to any one of claims 1 to 5 are implemented.