A multi-parameter identification method for electro-hydraulic position servo system
Through the reverse nonlinear sparrow search algorithm based on the punishment mechanism, combined with the error evaluation function, the efficiency and accuracy problems in the parameter identification of electro-hydraulic position servo system are solved, and more efficient parameter identification and model construction are achieved.
Patent Information
- Application Number
- CN202210855979.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-07-08
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Due to complex perturbation factors such as model uncertainty and parameter perturbation, the efficiency and accuracy of identifying parameters are low. Classical identification methods such as least squares method are difficult to effectively deal with higher-order systems, and the sparrow search algorithm is slow due to its convergence speed. It is easy to fall into local optimal problems, affecting its identification performance.
A reverse nonlinear sparrow search algorithm based on a punishment mechanism is proposed, which enriches the initial population diversity through reverse learning strategies, introduces nonlinear convergence factors, adaptive weight factors and gold cosine factors to balance local optimization and global search capabilities. And use the error evaluation function to identify and identify the results to obtain the accurate mathematical model of the electro-hydraulic position servo system.
It improves identification accuracy and efficiency, avoids algorithm precocious maturity and local optimal traps, enhances system operability, and realizes more efficient parameter identification and model construction.
Smart Images

Figure CN115185184B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of automatic control, and in particular relates to a multi-parameter identification method for an electro-hydraulic position servo system. Background Art
[0002] Electro-hydraulic position servo systems are widely used in the industrial field due to their good rapid response and strong load resistance, as well as their advantages of light weight and small size. However, electro-hydraulic servo systems have many complex disturbance factors such as model uncertainty and parameter perturbation, which will greatly affect the tracking and control accuracy of the system. Therefore, many experts and scholars are committed to studying the parameter identification method of electro-hydraulic position servo systems.
[0003] Classical identification methods mainly use the least square method to identify system parameters. Since the electro-hydraulic position servo system model has a high order, there are many parameters that need to be identified, and identification is difficult. The Sparrow Search Algorithm (SSA) is a heuristic swarm intelligence optimization algorithm based on biological behaviors such as sparrows foraging and avoiding predators. It is easy to implement, has a unique search model and excellent optimization capabilities, and can effectively handle global optimization and large-scale optimization problems. It is widely used in many fields such as multi-objective optimization, parameter identification, and parameter optimization. However, due to the problems of the initialization of the population position and the population position update method being too single, SSA has the defects of slow algorithm convergence speed, low convergence accuracy, and easy to fall into local optimality, which will affect the efficiency and accuracy of SSA identification parameters. Summary of the invention
[0004] In view of this, the present invention provides a multi-parameter identification method for an electro-hydraulic position servo system, so that the identification method has the advantages of high precision, low cost, high efficiency and strong operability.
[0005] The technical solution to implement the present invention is: a multi-parameter identification method for an electro-hydraulic position servo system, specifically comprising the following steps:
[0006] Step 1: Perform mathematical modeling based on the operating mechanism of the electro-hydraulic position servo system to determine the system model structure and parameters that need to be identified;
[0007] Step 2: A reverse nonlinear sparrow search algorithm based on a penalty mechanism is proposed. The reverse learning strategy is used to enrich the diversity of the initial population. Nonlinear factors are introduced in the position update process of the discoverer and the joiner, including nonlinear convergence factor, adaptive weight factor and golden sine and cosine factor, so as to balance the local optimization and global search capabilities. At the same time, a penalty mechanism is formulated according to the degree of danger discovered by individual sparrows to make the position distribution of the guards more reasonable.
[0008] Step 3: According to the acquired input and output sampling data of the electro-hydraulic position servo system, a reverse nonlinear sparrow search algorithm based on a penalty mechanism is used to identify the system parameters, and an error evaluation function is introduced to identify the identification results to obtain an accurate mathematical model of the electro-hydraulic position servo system.
[0009] Further, in step 1, the electro-hydraulic position servo system includes a workbench, a hydraulic cylinder, a displacement sensor, an electro-hydraulic servo valve, a servo amplifier, an electromagnetic reversing valve, a filter, an oil source and other structures, as shown in the attached manual. Figure 1 As shown in the figure, various parts of the system are modeled according to the operating mechanism of the electro-hydraulic position servo system.
[0010] Servo sensor model
[0011] The displacement sensor converts the displacement signal output by the hydraulic cylinder into an electrical signal. The displacement sensor equation is:
[0012] U f =K f x p
[0013] Where: K f is the displacement sensor gain coefficient; x p is the output displacement of the hydraulic cylinder.
[0014] Servo amplifier model
[0015] The servo amplifier of the servo valve consists of a signal conditioning circuit and a power amplifier circuit. Under the condition of ignoring the dynamic characteristics of the system, its output current is:
[0016] I=ΔuK a
[0017] Where: K a is the amplifier gain factor (A / V).
[0018] Transfer function of electro-hydraulic servo valve
[0019] Discussion on the second-order oscillation link of the servo valve:
[0020]
[0021] Where: K sv is the servo valve flow gain coefficient; ω sv is the natural frequency of the servo valve (rad / s); ζ sv is the servo valve damping ratio.
[0022] Basic equations of hydraulic power mechanism
[0023] The servo valve linearized flow equation, hydraulic cylinder continuity equation and hydraulic cylinder force balance equation of the electro-hydraulic position servo system are as follows:
[0024]
[0025] Where: m is the total mass of the piston and the load converted to the piston (kg); B p is the viscous damping coefficient (N·m / s); K is the equivalent spring stiffness of the hydraulic cylinder (N / m); F L is any external load acting on the piston (N), F f is the friction force acting on the piston (N), k q is the flow gain coefficient; k c is the flow rate and pressure coefficient, q L is the load flow; P L is the load pressure, C ip is the hydraulic cylinder internal leakage coefficient (m 3 / s·Pa -1 ), V t is the volume of the hydraulic cylinder chamber (m 3 ).
[0026] Parameters to be identified
[0027]
[0028] Where: k ce is the pressure coefficient of the total flow rate of the system (m 5 / N*m), k ce =k c +C ip .
[0029] Ignoring the related nonlinear factors, the structural block diagram of the electro-hydraulic position servo system model is as shown in the attached manual. Figure 2 As shown, the model transfer function formula is as follows:
[0030]
[0031] Where:
[0032] According to the transfer function, the electro-hydraulic position servo system is a fifth-order system. The fifth-order model structure is known, and the system can be closed-loop identified. The state space model of the fifth-order system is established. The state space model has good mathematical mapping capabilities and can solve high-order system and multivariable problems well. The system state variables are selected as:
[0033]
[0034] Establish the state space expression of the closed-loop system:
[0035]
[0036] in: B=[0 0 0 0 a1] T , C=[1 0 0 0 0], D=0; a1=K C , a3=2ω sv ω h (ξ h ω sv +ξ sv ω h ), a5=2(ξ sv ω sv +ξ h ω h ).
[0037] Therefore, the parameter vector to be identified is:
[0038] θ=[a1 a2 a3 a4 a5] T
[0039] Furthermore, in step 2, the reverse nonlinear sparrow search algorithm based on the penalty mechanism uses a reverse learning strategy to obtain a reverse population in the process of initializing the population. For example, in the d-dimensional search space, the individual is X i =(x1,x2,…,x d ), satisfying x i ∈[ub,lb] boundary condition, its inverse solution is Xx * =(x1 * ,x2 * ,…,x d * ), the mathematical expression is:
[0040]
[0041] Then select individuals with better fitness from the current population and the reverse population to construct the initialization population.
[0042] Furthermore, the penalty mechanism-based reverse nonlinear sparrow search algorithm introduces nonlinear factors into the position update formulas of the discoverer and joiner of the sparrow population, including nonlinear convergence factor, adaptive weight factor and golden sine and cosine factor, to improve the position update formulas of the discoverer and joiner.
[0043] In the traditional sparrow search algorithm, the finder position update formula is as follows:
[0044]
[0045] From the convergence function f(x)=exp[-x / (α·T)], it can be seen that as the value of x increases, the value of f(x) gradually decreases, and each dimension of the sparrow is shrinking. This linear change cannot adapt to the algorithm's optimization of complex multi-peak functions, resulting in a slow optimization speed of the algorithm, or even falling into a local optimal value. Therefore, the present invention introduces a nonlinear convergence factor, and the mathematical expression of the nonlinear convergence factor is as follows:
[0046]
[0047] Where: t is the current iteration number, f m is the initial value of the convergence factor. k is the control factor, which can control the attenuation amplitude, k∈[0,10].
[0048] Furthermore, an adaptive weight factor is introduced: in the early stage of algorithm iteration, a larger weight is given to make the population traverse the entire search space with a large step length, which is conducive to the algorithm to quickly search for the global optimal position; in the later stage of algorithm iteration, a smaller weight is given to help the algorithm explore the optimal position with a small step length, improving the convergence accuracy of the algorithm. The mathematical model of the adaptive weight factor is as follows:
[0049]
[0050] Where: δ1, δ2, δ3, ρ1, ρ2, ρ3 are constant coefficients, and τ is the specified number of iterations.
[0051] Furthermore, the update formula of the finder position in the reverse nonlinear sparrow search algorithm based on the penalty mechanism is:
[0052]
[0053] Furthermore, the golden sine algorithm (golden-SA) is a process of scanning the unit circle simulation algorithm search space through the sine function. Its mathematical expression is described as follows:
[0054]
[0055] Where: R1 and R2 are random numbers in [0,2π] and [0,π], respectively, indicating the distance and direction of movement of the next generation of individuals. is the optimal individual position for the tth iteration. x1 and x2 are the golden section coefficients, which are used to narrow the search space and guide the individual to converge to the optimal value. x1=a(1-γ)+bγ,x2=aγ+b(1-γ), γ is the golden ratio, which is approximately 0.618033, and a and b are the search intervals.
[0056] Furthermore, according to the updated position status of the discoverer, a sine and cosine search is performed to obtain the optimal feasible solution. The mathematical expression of the joiner position update based on the sine and cosine search strategy is as follows:
[0057]
[0058] Where: x3 is a random number uniformly distributed on [0,1].
[0059] Furthermore, according to the degree to which individual sparrows detect danger, a penalty mechanism is formulated to improve the location distribution of the sentinels. Simply placing the sentinels at the optimal solution will result in a waste of resources. Here, the distance of the sparrows at the edge of the population from the safe area is considered for processing.
[0060]
[0061] Where: f1 is the individual fitness value of the sparrow using the jump mechanism, and f2 is the individual fitness value of the sparrow using the penalty mechanism.
[0062] The more the sparrows leave the safe area, the closer they are pulled back. The smaller the safe zone is, the closer it will be when it is pulled back. It can ensure that the processed position is within the preset range.
[0063] Furthermore, in step 3, an error evaluation function is introduced to identify the identification results, and the process of system parameter identification is transformed into a function fitting process. Assume that in the time domain, the relationship between the input and output of the system is as follows:
[0064] y(kT s )=f[u(kT s )],k=1,2,…,M
[0065] Where: M is the number of sampling points, T s is the sampling period.
[0066] When M sets of input and output data y(kT s ) and u(kT s ), then a known function f that approximately matches f is obtained through the sparrow search algorithm. g Taking into account the nonlinear factors that are difficult to predict in the actual process, the actual identification model can be expressed as:
[0067] y(kT s ) = f g [u(kT s )]+e(KT s ),k=1,2,…,M
[0068] In the formula: e(kT s) is the residual value caused by actual factors.
[0069] The error evaluation function is defined as:
[0070]
[0071] In general, the above technical solutions conceived by the present invention can achieve the following beneficial results compared with the prior art:
[0072] (1) First, a model is constructed according to the operating mechanism of the electro-hydraulic position servo system, and then the parameters and parameter ranges to be identified are determined. Compared with directly identifying the system, this identification method is more accurate, more efficient, and more operational;
[0073] (2) In the population position initialization stage, a reverse learning strategy is used to generate a reverse population, which can not only greatly improve the quality of the solution, but also increase the exploratory nature of individuals in the population, broaden the population search area, improve the population diversity, and effectively prevent the algorithm from being premature.
[0074] (3) Adding nonlinear factors to the position update formulas of sparrow finders and joiners can not only further improve the convergence accuracy and optimization effect of the algorithm and accelerate the convergence speed, but also balance the local development and global search capabilities;
[0075] (4) According to the degree to which individual sparrows detect danger, a penalty mechanism is formulated to improve the distribution of the positions of the sentinels, which can prevent the algorithm from falling into the local optimum and further improve the convergence accuracy and efficiency of each iteration;
[0076] (5) Using the defined error evaluation function to identify the identification results can make the identification results more accurate and identify the system parameters faster. BRIEF DESCRIPTION OF THE DRAWINGS
[0077] Figure 1 Schematic diagram of electro-hydraulic position servo system
[0078] Figure 2 Electro-hydraulic position servo system structure diagram
[0079] Figure 3 Flow chart of system parameter identification based on the reverse nonlinear sparrow search algorithm of penalty mechanism
[0080] Figure 4 Comparison of the identification system response curve and the input step target curve
[0081] Figure 5 Comparison of the identification system response curve and the input sinusoidal target curve
[0082] Figure 6 A comparison of the response curve of the identification system and the input sinusoidal target curve DETAILED DESCRIPTION
[0083] In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the purpose, technical solution and advantages of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. The following examples are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.
[0084] Step 1. Model the system according to the operating mechanism of the electro-hydraulic position servo system, obtain the system model structure and the parameters that need to be identified, and determine the parameter variation range, as shown in the following formula:
[0085]
[0086] Parameters to be identified:
[0087] θ=[a1 a2 a3 a4 a5] T , where a1=K C , a3=2ω sv ω h (ξ h ω sv +ξ sv ω h ), a5=2(ξ sv ω sv +ξ h ω h );
[0088] Step 2. Introduce an error evaluation function to identify the identification results, transform the system parameter identification into an optimization problem, and define the error evaluation function as:
[0089]
[0090] Where: M is the number of sampling points, T s is the sampling period, e(kT s ) is the residual value caused by actual factors, u(kT s ) is the input sampling data, y(kT s ) is the output sampling data;
[0091] Step 3. Initialize population parameters, including the number of individuals N as 30, the maximum number of iterations Max_iter as 500, the dimension d, the population search boundaries lb, ub, etc., and set related parameters, such as the discoverer ratio PD and the joiner ratio SD, the warning threshold R2, etc.;
[0092] Step 4. In the population position initialization stage, a reverse learning strategy is used to generate a reverse population, and individuals with higher fitness priority levels are selected from the current population and the reverse population to construct an initialization population, and the initial position information of the sparrow individuals is established as:
[0093] Step 5. Calculate the fitness value of each sparrow, find out the current optimal fitness value and the worst fitness value, and the corresponding position;
[0094] Step 6. Select some sparrow individuals with better fitness from the sparrow population as discoverers, and the remaining sparrows as joiners. Introduce nonlinear convergence factor f, adaptive weight factor ω and golden sine and cosine factors, take dimension d = 30, and set the adaptive weight factors to: δ1 = 0.35, δ2 = 250, δ3 = 1.5, ρ1 = 0.09, ρ2 = 0.015, ρ3 = 0.15, τ = 14, a = π, b = -π, and perform position update operations on discoverers and joiners according to the improved discoverer and joiner position update formula;
[0095] Step 7. Randomly select some sparrows from the sparrows as sentinels, and perform position update operations according to the improved sentinel position update formula based on whether they are threatened. For the threatened sparrow individuals, the penalty mechanism and jump mechanism are respectively operated, and the greedy selection rule is introduced to determine the update method by comparing the fitness values of the two positions;
[0096] Step 8. Determine whether the entire program has reached the end condition. If not, jump to step 5 and restart the iteration, and the number of iterations t=t+1; otherwise, determine that the loop is over and output the optimal result;
[0097] Step 9. Perform data acquisition on the test bench with a sampling time of 0.001s, and perform data preprocessing on the data obtained from the final sampling. Collect the system step input and sinusoidal input signals. Repeat the comparative test of system identification 20 times. Take the average value of the results of 20 runs. Take the step and sinusoidal input signals of the identified system as input signals, and compare the response curve with the input target curve. Figure 4 , 5 are step input and sinusoidal input respectively, Figure 6 This is a local enlarged view of the sinusoidal input.
[0098] It can be seen from the figure that no matter it is the identification result curve after sampling step input or sinusoidal input data, the identified system response curve has a certain error with the input target curve, but it can reflect the characteristics of the input signal and has a certain degree of accuracy.
[0099] The above implementation modes are only used to illustrate the present invention, but not to limit the present invention. Ordinary technicians in the relevant technical field may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all equivalent technical solutions also belong to the scope of the present invention. The protection scope of the present invention should be defined by the claims.
Claims
1. A multi-parameter identification method for an electro-hydraulic position servo system, characterized in that: The following steps are involved: Step 1: Perform mathematical modeling based on the operating mechanism of the electro-hydraulic position servo system to determine the system model structure and parameters that need to be identified; Step 2: A reverse nonlinear sparrow search algorithm based on a penalty mechanism is proposed. The reverse learning strategy is used to enrich the diversity of the initial population. Nonlinear factors are introduced in the position update process of the discoverer and the joiner, including nonlinear convergence factor, adaptive weight factor and golden sine and cosine factor, so as to balance the local optimization and global search capabilities. At the same time, a penalty mechanism is formulated according to the degree of danger discovered by individual sparrows to make the position distribution of the guards more reasonable. The characteristics are: Step 2-1 Initialization of sparrow population position The reverse learning strategy is used to improve the algorithm's search ability. Specifically, in the d-dimensional search space, the individual is X i =(x1,x2,…,x d ), satisfying x i ∈[ub,lb] boundary condition, the inverse solution is X i * =(x1 * ,*2 * ,…,x d * ), the mathematical expression is: Then, individuals with higher fitness priority levels are selected from the current population and the reverse population to construct the initialization population; Step 2-2 Finder Location Update The nonlinear convergence factor and adaptive weight factor are introduced into the discoverer position update formula, specifically: The nonlinear convergence factor f is used to replace the convergence function f(x)=exp[-x / (α·T)] in the position update formula. The mathematical expression of the nonlinear convergence factor is as follows: Where: t is the current iteration number, f m is the initial value of the convergence factor, k is the control factor, which can control the attenuation amplitude, and k∈[0,10]; The mathematical model of adaptive weight factor is as follows: Where: δ1, δ2, δ3, ρ1, ρ2, ρ3 are constant coefficients, and τ is the specified number of iterations; After adding the nonlinear convergence factor and adaptive weight factor, the finder position update formula is: Step 2-3 Joiner location update The golden sine and cosine factors are introduced into the joiner position update formula, and its mathematical expression is described as follows: Where: R1 and R2 are random numbers in [0,2π] and [0,π], respectively, indicating the distance and direction of movement of the next generation of individuals. is the optimal individual position for the tth iteration; x1 and x2 are the golden section coefficients, which are used to guide the individual to converge to the optimal value; x1 = a(1-γ) + bγ, x2 = aγ + b(1-γ), γ is the golden cutting ratio, which is approximately 0.618033, and a and b are the search intervals; According to the updated position status of the discoverer, a sine-cosine search is performed to obtain the optimal feasible solution; the mathematical expression of the joiner position update based on the sine-cosine search strategy is as follows: Where: x3 is a random number uniformly distributed on [0,1]; Step 2-4: Update the location of the Sentinel When a sparrow is at the edge of the population and is threatened, a punishment mechanism is formulated according to the degree to which the sparrow detects danger, and the position of the sentinel is updated. Specifically: Where: f1 is the individual fitness value of the sparrow using the jump mechanism, and f2 is the individual fitness value of the sparrow using the penalty mechanism; Step 3: According to the acquired input and output sampling data of the electro-hydraulic position servo system, a reverse nonlinear sparrow search algorithm based on a penalty mechanism is used to identify the system parameters, and an error evaluation function is introduced to identify the identification results to obtain an accurate mathematical model of the electro-hydraulic position servo system.
2. The multi-parameter identification method of an electro-hydraulic position servo system according to claim 1, characterized in that: Step 3: Based on the obtained system model structure, parameters to be identified, and input and output data, an error evaluation function is introduced to identify the identification results, specifically: The electro-hydraulic position servo system model is: Parameter to be identified: θ = [a1 a2 a3 a4 a5] T , where a1 = K C , a3 = 2ω sv ω h (ξ h ω sv + ξ sv ω h ), a5 = 2(ξ sv ω sv + ξ h ω h ); Based on the sampled system input and output data, it is assumed that in the time domain, the relationship between the system input and output is as follows: y(kT s )=f[u(kT s )],k=1,2,…,M Where: M is the number of sampling points, T s is the sampling period; When M sets of input and output data y(kT s ) and u(kT s ), a known function f that approximately matches f is obtained through a reverse nonlinear sparrow search algorithm based on a penalty mechanism. g ; Considering the nonlinear factors that are difficult to predict in the actual process, the actual identification model can be expressed as: y(kT s )=f g [u(kT s )]+e(kT s ),k=1,2,…,M In the formula: e(kT s ) is the residual value caused by actual factors; The error evaluation function is defined as:
