Snake-vulture hybrid algorithm-based active disturbance rejection fault-tolerant control method and system

Abstract

The application discloses a kind of based on snake hawk hybrid algorithm's self-disturbance rejection fault-tolerant control method and system, and relates to mechanical control technical field.The position of firefly is calculated formula is integrated into the attack prey formula of snake hawk algorithm;The formula of fitness function including the dynamic decision domain radius of two random candidate solutions of snake hawk hybrid algorithm is improved in firefly algorithm;The dynamic decision domain radius in improved firefly algorithm is integrated into the escape strategy of snake hawk algorithm, the global search capability of firefly algorithm in later period is strengthened, and the algorithm is avoided to fall into local optimum quickly.Snake hawk hybrid algorithm is used to optimize self-disturbance rejection fault-tolerant control method and Elman neuron network, and the best neuron network parameter is obtained, which is converted based on real-time operation parameter for calculation, to ensure that it can reliably and stably work.

Description

Technical Field

[0001] This invention relates to the field of mechanical control technology, and in particular to an active disturbance rejection and fault-tolerant control method and system based on the guilder hybrid algorithm. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In contemporary society, machinery aids development in various fields. In agricultural production, large combine harvesters operate efficiently, and seeders sow precisely, improving production efficiency and crop yields, thus promoting agricultural modernization. Construction machinery is equally crucial in engineering projects. During the construction of roads, bridges, and buildings, excavators, loaders, and cranes work together to accelerate progress and ensure construction quality. However, machinery operation faces various challenges. Taking agricultural machinery as an example, due to the unique characteristics of the field environment, when agricultural machinery is working, tractors encountering tree roots or stones during plowing can cause a rapid increase in plowing force. Even worse, foreign objects such as straw can get caught in the working device, rapidly increasing the load on the device, rendering the machine inoperable, or even causing safety accidents, threatening production and safety.

[0004] To address the aforementioned issues, existing technologies improve controllers, such as by adding active disturbance rejection mechanisms (ADRMs) to cope with complex environments. However, due to the uncontrollability of the working environment, existing ADRMs have poor adaptability to surrounding disturbances and lack fault tolerance, thus failing to guarantee stable mechanical operation. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide an active disturbance rejection and fault-tolerant control method and system based on the Scorpion Hybrid Algorithm. This method constructs a fitness function for active disturbance rejection and fault-tolerant control of the controller, and utilizes the Scorpion Hybrid Algorithm based on the Scorpion Algorithm and the Firefly Algorithm to optimize the neural network parameters, thereby ensuring the stability of agricultural or engineering machinery operations.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0007] The first aspect of this invention provides an active disturbance rejection and fault-tolerant control method based on the Scorpion Eagle hybrid algorithm, comprising the following steps:

[0008] The system acquires the operating parameters and measured force data of the machine to be controlled. Based on the operating parameters and the working principle of the operating device, it calculates the operating force data. The calculated operating force data is subtracted from the measured force data to obtain the difference. The difference is used to construct a fitness function, and the neural network is trained based on the fitness function.

[0009] By introducing the firefly algorithm into the scorpion algorithm, a scorpion hybrid algorithm is obtained. The formula for calculating the firefly's position is incorporated into the prey attack formula of the scorpion algorithm. The fitness function formula of the scorpion hybrid algorithm is constructed to improve the dynamic decision domain radius in the firefly algorithm. The improved dynamic decision domain radius of the firefly algorithm is then incorporated into the escape strategy of the scorpion algorithm.

[0010] The neural network was iteratively optimized using the Scorpion Hybrid Algorithm to obtain a neural network prediction model for the job performance difference.

[0011] The trained neural network prediction model for the difference in working force is transformed into a multi-segment regression model for the difference in working force. The multi-segment regression model is used to calculate the actual difference in the real-time working parameters collected. The working force calculated in real time based on the working principle of the working device is combined with the actual difference to form an active disturbance rejection and fault tolerance strategy, which is then input into the controller to control the machine to be controlled.

[0012] A second aspect of the present invention provides an active disturbance rejection and fault-tolerant control system based on the Scorpion Sparrow hybrid algorithm, comprising:

[0013] The model training module is configured to acquire the operating parameters and measured force data of the machine to be controlled, calculate the operating force data based on the operating parameters and the working principle of the operating device, subtract the calculated operating force data from the measured force data to obtain the difference, construct a fitness function with the difference as the target, and train the neural network based on the fitness function.

[0014] The algorithm improvement module is configured to introduce the firefly algorithm into the guilder algorithm to obtain the guilder hybrid algorithm. Specifically, the formula for calculating the firefly's position is incorporated into the prey attack formula of the guilder algorithm. The fitness function formula of the guilder hybrid algorithm is constructed to improve the dynamic decision domain radius in the firefly algorithm. The improved dynamic decision domain radius of the firefly algorithm is incorporated into the guilder algorithm's escape strategy. The guilder hybrid algorithm is used to iteratively optimize the neural network to obtain the job performance difference neural network prediction model.

[0015] The mechanical control module is configured to transform the trained working force difference neural network prediction model into a multi-segment working force difference regression model. The multi-segment working force difference regression model is used to calculate the actual difference based on the collected real-time working parameters. The working force calculated in real time based on the working principle of the working device is combined with the actual difference to form an active disturbance rejection and fault tolerance strategy, which is then input into the controller to control the machine.

[0016] A third aspect of the present invention provides a medium having a program stored thereon, which, when executed by a processor, implements the steps of the active disturbance rejection and fault-tolerant control method based on the Osprey hybrid algorithm as described in the first aspect of the present invention.

[0017] A fourth aspect of the present invention provides an apparatus including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the active disturbance rejection and fault-tolerant control method based on the guilder hybrid algorithm as described in the first aspect of the present invention.

[0018] The above one or more technical solutions have the following beneficial effects:

[0019] This invention discloses an active disturbance rejection and fault-tolerant control method and system based on the Scorpion Hybrid Algorithm. It designs an active disturbance rejection and fault-tolerant controller based on the Scorpion Algorithm and the Firefly Algorithm. Specific improvements include: incorporating the firefly's position calculation formula into the Scorpion Algorithm's prey attack formula; constructing a formula that includes the fitness function of two random candidate solutions from the Scorpion Hybrid Algorithm to improve the dynamic decision domain radius in the Firefly Algorithm; and integrating the improved dynamic decision domain radius from the Firefly Algorithm into the Scorpion Algorithm's escape strategy, thereby enhancing the Firefly Algorithm's global search capability in the later stages and preventing the algorithm from quickly falling into local optima.

[0020] In this invention, an active disturbance rejection and fault-tolerant control method based on the Scorpion Hybrid Algorithm is used to optimize the Elman neural network, obtaining the optimal neural network parameters. Based on this neural network model, active disturbance rejection and fault-tolerant control is applied to the machinery, enabling it to respond more accurately to external interference and internal faults during operation, reducing operational deviations, maintaining stable operation, thereby improving work quality and efficiency, reducing downtime and maintenance costs caused by faults, and ensuring reliable and stable operation.

[0021] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0022] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0023] Figure 1 This is a flowchart of the active disturbance rejection and fault-tolerant control method based on the Scorpion Eagle hybrid algorithm in Embodiment 1 of the present invention. Detailed Implementation

[0024] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0025] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0026] Example 1:

[0027] Embodiment 1 of this invention provides an active disturbance rejection and fault-tolerant control method based on the Scorpion Eagle hybrid algorithm, such as... Figure 1 As shown, it includes the following steps:

[0028] Step 1: Obtain the operating parameters and measured force data of the machine to be controlled.

[0029] In one specific implementation, the operating parameters include travel speed, operating device speed, operating device upward acceleration, operating device downward acceleration, output torque and speed of the power unit of agricultural machinery or engineering machinery, and the rate of change of output torque and speed of the power unit of agricultural machinery or engineering machinery.

[0030] Data on the measured force is collected using a specially equipped force sensor.

[0031] Step 2: Calculate the working force data based on the working parameters and the working principle of the working device. Subtract the calculated working force data from the measured force data to obtain the difference. Use the difference as the target to construct a fitness function and train the neural network based on the fitness function.

[0032] In one specific implementation, a large amount of operational parameters of agricultural or construction machinery are collected, such as travel speed, operating device speed, operating device acceleration (ascent and descent), output torque and speed of the power unit, and the rate of change of output torque and speed of the power unit. Based on these operational parameters and the working principle of the operating device, the operating force data is calculated. However, since the existing data calculated based on the operating principle of the operating device differs from the actual required data and lacks fault tolerance, the calculated operating force data is subtracted from the measured force data to obtain the difference. This difference is used to construct a fitness function, and a neural network is trained based on the fitness function to form a self-disturbance rejection and fault-tolerant strategy to control the machinery, ensuring stable operation of the agricultural or construction machinery.

[0033] Specifically, based on the operating parameters and the working principle of the operating device, the operating force data is calculated as follows: The data of the working force measured by the force sensor is as follows Data on actual working capacity Data on calculated working force Subtracting the two values ​​yields the difference. Since the difference can be positive or negative, two sample sets are constructed based on this difference. These sample sets include a positive sample set (composed of positive differences) and a negative sample set (composed of negative differences). Therefore, the sample sets consist of the job operation parameters and their corresponding job force differences. A fitness function is then constructed using these job force differences. Among them, f e t (·) represents the objective function value corresponding to the solution, where represents the fitness function. The neural network is trained using two sample sets based on the fitness function.

[0034] It is important to note that the training set consists of two sets: a positive sample set and a negative sample set. Since negative differences cannot be used for training, the differences in the training set formed by the negative sample set are also trained using positive values ​​in an absolute value manner. Therefore, there are two resulting neural network prediction models for the work capacity difference. The work capacity difference prediction model trained on the negative sample set needs to have a negative sign added to the actual prediction value. Furthermore, after converting the trained work capacity difference neural network prediction model into a multi-segment work capacity difference regression model, the actual difference calculated based on the collected real-time work operation parameters is positive, while the actual difference calculated based on the negative sample set needs to have a negative sign added.

[0035] In this embodiment, the working principle of the working device is derived from the working conditions of different machines, which is the prior art in this field and will not be described in detail here.

[0036] Step 3: By introducing the firefly algorithm into the guildhrone algorithm, the guildhrone hybrid algorithm is obtained.

[0037] Specifically, the formula for calculating the firefly's position is incorporated into the prey attack formula of the Scorpion Algorithm, and a formula for improving the dynamic decision domain radius in the firefly algorithm is constructed, which includes the fitness function of two random candidate solutions from the Scorpion hybrid algorithm. The improved dynamic decision domain radius in the firefly algorithm is then incorporated into the escape strategy of the Scorpion Algorithm.

[0038] The detailed steps are as follows:

[0039] 1. Firefly Algorithm:

[0040] The Firefly Algorithm is an optimization algorithm based on the simulation of the behavior of fireflies. This algorithm searches for the optimal solution in the solution space by simulating the behavior of fireflies in their environment.

[0041] In the GSO algorithm, each firefly represents a solution and has a brightness value that represents the fitness of that solution. Fireflies interact with each other by emitting light and sensing light. Each firefly emits luciferin, which is proportional to its brightness value. Other fireflies sense and follow the light source based on the magnitude of the luciferin value, thus realizing the solution search process.

[0042] During the search process, each firefly updates its position and brightness based on its own and the brightness of surrounding fireflies, thus optimizing the solution. The closer the fireflies are, the stronger their interaction; the brighter a firefly is, the greater its influence. Based on these rules, the GSO algorithm continuously updates the fireflies' positions and brightness until it finds the optimal solution.

[0043] The specific algorithm is as follows:

[0044] (1) Algorithm initialization:

[0045] Initialize and place n y Given a firefly, assign an initial luciferin l0 to each firefly, set a dynamic decision region r0, and provide an initial step size s for the firefly algorithm. y Set the threshold n of the neighborhood. t Luciferin disappearance rate ρ, Luciferin renewal rate γ, dynamic decision domain renewal rate β, firefly perception domain r s Number of iterations M.

[0046] (2) Update fluorescein:

[0047] l k (t)=(1-ρ)l k (t-1)+γF(x k (t)).

[0048] In the formula: l k (t) represents the luciferin intensity of firefly k in generation t; F(x) k (t) represents the fitness function represented by the position of firefly k in generation t; x k (t) represents the position of firefly k in generation t.

[0049] (3) Update the neighbors of firefly k:

[0050]

[0051] Where: N k (t) represents the set of the k neighbors of the firefly in generation t; The dynamic decision domain of firefly k in generation t; l j (t) represents the luciferin intensity of firefly j in generation t; firefly k and firefly j are different fireflies.

[0052] (4) Use the roulette wheel method to determine the firefly's next movement direction:

[0053] j = max(p k )

[0054]

[0055] In the formula: p kj p represents the probability of firefly k transitioning to firefly j, and its value is the ratio of the difference in luciferin levels between the two fireflies to the sum of the luciferin differences of firefly k's neighbors; k This indicates that firefly k moves towards its N. k The vector consisting of the transition probabilities of moving (t) neighbors.

[0056] (5) Update the firefly locations:

[0057]

[0058] This formula indicates that the position of firefly k in generation t+1 is determined by the current position in generation t via a fixed step size s. y It moves towards the position of firefly j in the manner described by x. k (t+1) represents the position of firefly k in generation t+1; x i (t) represents the position of firefly i in generation t, where firefly i is a different firefly from firefly k and firefly j; x j (t) represents the position of firefly j in time t.

[0059] (6) Update the radius of the dynamic decision domain:

[0060]

[0061] The above formula represents the radius of the decision domain of firefly k in generation t+1. It is based on the size of the decision domain radius of its previous round. and the current distance to it is less than or equal to The number of fireflies N k The difference in (t) is used to determine this. If N k (t) is greater than the total number of iterations n in the current iteration rounds. t If the number of fireflies in the neighborhood of firefly k is saturated, then the decision domain radius needs to be reduced to enhance the finer search; conversely, if N k (t) is less than n t This indicates that there is still enough space to search within the neighborhood of firefly k, and its decision domain radius can be appropriately expanded to increase the search speed. In the update formula, The purpose of this is to ensure that the decision domain radius does not become negative, while the purpose of the entire dynamic decision domain formula is to limit the maximum value of the decision domain radius to prevent the decision domain radius from becoming too large and causing the search range to expand indefinitely.

[0062] 2. Scorpion Algorithm:

[0063] The secretary bird algorithm simulates the hunting behavior of secretary birds. During the hunt, the secretary bird searches for prey within a certain area, and upon finding prey, it adopts different strategies based on the prey's state, including searching, attacking, and fleeing.

[0064] Hunting: Simulate the secretary bird's search for prey, using its incredibly keen eyesight to track and find snakes hidden in the grass.

[0065] Attacking prey: The secretary bird does not rush to attack. Instead, it uses agile footwork and mobility to observe the snake, and by keenly judging the snake's movements, it gradually provokes and wears down the snake's stamina. When the snake is exhausted, it launches an attack.

[0066] Escape strategy: The secretary bird has a large number of natural enemies. When threatened, the secretary bird will flee or disguise itself to avoid attacks from its natural enemies.

[0067] The specific algorithm is as follows:

[0068] (1) Initialization phase:

[0069] x p,q =lb q +r×(ub q -lb q ), p=1,2,...,M, q=1,2,...,Dim.

[0070] In the formula: x p,qThis represents the current position of the p-th q-th guild bird; r represents a random number between 0 and 1; lb j and ub j Here, M represents the lower and upper bounds of the search space, respectively; M represents the number of iterations; and Dim represents the dimension.

[0071] (2) Consume prey:

[0072] RB = randn(1,Dim).

[0073]

[0074] In the formula: randn(1,Dim) represents an array of dimension 1×Dim randomly generated from a standard normal distribution (mean is 0, standard deviation is 1); This represents the current position of the q-th dimension of the necromancer in the first phase, which is the prey-attacking phase. This represents the optimal position of the guilder at the t-th iteration; t represents the current iteration number; T represents the maximum iteration number; RB is a 1×Dim array randomly generated from a standard normal distribution.

[0075] (3) Attacking prey:

[0076]

[0077] RL = 0.5 × Levy(Dim),

[0078]

[0079] In the formula: s e Let n be the initial step size of the Scorpion Algorithm, a fixed constant of 0.01. e η is the number of secretary birds; η is a fixed constant of 1.5; u and v are random numbers in the interval [0,1]; Γ represents the gamma function; RL represents weighted Levy(Dim) to improve the optimization accuracy of the algorithm; Levy(Dim) is the Levy flight parameter, and Dim is the dimension; σ is an intermediate calculation variable that provides data support for determining the Levy flight parameter throughout the calculation process, thus affecting the calculation of the new position when attacking prey.

[0080] (4) Escape phase:

[0081]

[0082] In the formula: This represents the current position of the q-th dimension of the guild vulture during the second phase, i.e., the escape phase. This represents the optimal position of the guilder at the t-th iteration.

[0083] 3. Scorpion Hybrid Algorithm

[0084] Based on the aforementioned Scorpion Algorithm and Firefly Algorithm, in order to achieve better self-disturbance rejection and fault-tolerant control, this embodiment incorporates the firefly's position calculation formula into the Scorpion Algorithm's prey attack formula; constructs a formula that includes the fitness function of two random candidate solutions from the Scorpion hybrid algorithm to improve the dynamic decision domain radius in the firefly algorithm; and adopts the improved dynamic decision domain radius from the firefly algorithm into the Scorpion Algorithm's escape strategy, thereby strengthening the firefly algorithm's global search capability in the later stages and preventing the algorithm from quickly getting trapped in local optima.

[0085] First, based on the firefly transfer probability p kj The optimization process for determining the value of the target parameter involves deciding whether to use the original attack prey formula of the Scorpion Algorithm or the improved attack prey formula. Specifically, when the firefly transfer probability is greater than or equal to a set threshold, the Scorpion Algorithm and the Firefly Algorithm are combined, and ρ is added to the Firefly Algorithm. t The algorithm is modified to obtain an improved prey attack formula. When the firefly transfer probability is less than or equal to a set threshold, the original prey attack formula of the algorithm is used. In this embodiment, the threshold is set to 0.5, and the formula is as follows:

[0086]

[0087] During the optimization process, a counter `count` is introduced, initially set to 0. ρ t The calculation formula is: the optimal position of the guilder at the (t-1)th iteration. The optimal position of the guilder at iteration t. ratio ρ t The closer the value is to 1, the more similar the results of local optimization.

[0088] If |ρ t -1|<δ, take δ=0.01, then count=count+1; otherwise count=0.

[0089] When the count accumulates to a certain number of times T, it indicates that the results are very similar, resulting in a local optimum. At this point, it is necessary to increase the dynamic policy domain radius. This embodiment adjusts the radius in the firefly algorithm by setting an adjustment coefficient C.

[0090]

[0091] In the formula: t is the current iteration number; T is the maximum iteration number; and It is the position of the random candidate solution in the first stage of selection; f e t (·) represents the objective function value corresponding to the solution. In this formula, This reflects the iterative process; as the number of iterations t gradually approaches the maximum number of iterations T, It will gradually increase from 0 to 1. However, C cannot expand indefinitely. When C ≥ 100, let C = 100.

[0092] The difference in objective function values ​​between two candidate solutions is used to measure the degree of difference between solutions within the current search region. A large difference in objective function values ​​indicates high diversity of solutions within the region, suggesting that better solutions may remain undiscovered. Conversely, a small difference indicates similarity among solutions within the region, requiring a larger radius to escape local optima.

[0093] The formula for updating the radius of the dynamic decision domain is:

[0094]

[0095] After introducing the adjustment factor C, the formula becomes

[0096]

[0097] As C increases, C×β(n) t -|N k The value of (t)|) will also increase, and at this time, the increase of C will increase the radius of the decision domain. The increase is larger, thus expanding the search range and allowing more local optima to be obtained; r s This represents the sensory domain of a firefly.

[0098] Furthermore, the variation of the firefly's decision domain radius is incorporated into the escape strategy of the Scorpion algorithm to escape local optima. The formula is as follows:

[0099]

[0100] In the formula: This represents the radius of the decision domain of firefly k in generation t+1. Compared to the size of the decision domain radius in the previous round The ratio of .

[0101] This leads to the guilder hybrid algorithm.

[0102] The strategy is executed NZ times until NZ > Tt, where Tt is the number of iterations. In this embodiment, Tt = 30 is taken. At this point, the optimal value is found, thus obtaining the Scorpion-Bird hybrid algorithm.

[0103] Step 4: Iteratively optimize the neural network using the Scorpion Hybrid Algorithm to obtain the working power difference neural network prediction model.

[0104] In this embodiment, an Elman neural network is used. An Elman neural network generally consists of an input layer, hidden layers, a receiving layer, and an output layer. The input layer units transmit signals, the output layer units perform linear weighting, the hidden layer units are generally nonlinear activation functions, and the receiving layer receives feedback signals from the hidden layer to remember the output values ​​of the hidden layer units from the previous moment; it can be considered a time delay operator. The Elman neural network can better adapt to the dynamically changing working environment and data during machine operation, accurately simulate the complex laws governing the operation of the mechanical controller, and thus achieve more accurate disturbance rejection and fault-tolerant control.

[0105] In one specific implementation, during the training process, based on a large amount of historical data, the parameters of the neural network are continuously adjusted according to the error, gradually reducing the error, thereby obtaining a better Elman neural network and improving the accuracy and control performance of the model.

[0106] The Elman neural network learning process is as follows:

[0107] Let the external input of the network be u(d-1), the output be y(d), and the output of the hidden layer be x(d). Then, the following nonlinear state-space expression exists:

[0108]

[0109] X c (d) = X(d-1)

[0110]

[0111] In the formula: These are the connection weight matrices from the receiving layer to the hidden layer, from the input layer to the hidden layer, and from the hidden layer to the output layer, respectively; f(·) is the transfer function of the hidden layer neurons; g(·) is the transfer function of the output neurons, which is a linear combination of the outputs of the intermediate layers; X c (d) represents the output of the receiving layer at time d, and X(d-1) represents the output of the hidden layer at time d-1.

[0112] To achieve better self-disturbance rejection characteristics, the Osprey hybrid algorithm in this embodiment is used to improve the Elman neural network. Specifically, the fitness function is used as the objective, and the operating parameters such as driving speed, operating device speed, operating device upward acceleration, operating device downward acceleration, output torque and speed of agricultural machinery or construction machinery power device, and the rate of change of output torque and speed of agricultural machinery or construction machinery power device are used as inputs. The absolute value of the operating force difference is used as the output. The connection weight matrix of the Elman neural network is iteratively optimized using the Osprey hybrid algorithm to obtain the operating force difference neural network prediction model.

[0113] Step 5: Since the neural network model cannot be directly applied to the controller, the trained working force difference neural network prediction model is transformed into a multi-segment working force difference regression model. The multi-segment working force difference regression model is used to calculate the collected real-time working parameters and calculate the actual difference. The working force calculated in real time according to the working principle of the working device is combined with the actual difference to form an active disturbance rejection fault tolerance strategy, which is then input into the controller to control the machine to be controlled.

[0114] It should be noted that this embodiment uses agricultural machinery and engineering machinery as examples. In other possible embodiments, the active disturbance rejection fault-tolerant control method and controller of this embodiment can also be applied to other machinery to improve the stability of control.

[0115] Example 2:

[0116] Embodiment 2 of the present invention provides an active disturbance rejection and fault-tolerant control system based on the Scorpion Sparrow hybrid algorithm, comprising:

[0117] The model training module is configured to acquire the operating parameters and measured force data of the machine to be controlled, calculate the operating force data based on the operating parameters and the working principle of the operating device, subtract the calculated operating force data from the measured force data to obtain the difference, construct a fitness function with the difference as the target, and train the neural network based on the fitness function.

[0118] The algorithm improvement module is configured to introduce the firefly algorithm into the guilder algorithm to obtain the guilder hybrid algorithm. Specifically, the formula for calculating the firefly's position is incorporated into the prey attack formula of the guilder algorithm. The fitness function formula of the guilder hybrid algorithm is constructed to improve the dynamic decision domain radius in the firefly algorithm. The improved dynamic decision domain radius of the firefly algorithm is incorporated into the guilder algorithm's escape strategy. The guilder hybrid algorithm is used to iteratively optimize the neural network to obtain the job performance difference neural network prediction model.

[0119] The mechanical control module is configured to transform the trained working force difference neural network prediction model into a multi-segment working force difference regression model. The multi-segment working force difference regression model is used to calculate the actual difference based on the collected real-time working parameters. The working force calculated in real time based on the working principle of the working device is combined with the actual difference to form an active disturbance rejection and fault tolerance strategy, which is then input into the controller to control the machine.

[0120] Example 3:

[0121] Embodiment 3 of the present invention provides a medium on which a program is stored. When the program is executed by a processor, it implements the steps in the active disturbance rejection and fault-tolerant control method based on the Scorpion Sparrow hybrid algorithm as described in Embodiment 1 of the present invention.

[0122] Example 4:

[0123] Embodiment 4 of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the active disturbance rejection and fault tolerance control method based on the serpentine hybrid algorithm described in Embodiment 1 of the present invention.

[0124] The steps and methods involved in Examples 2, 3 and 4 above correspond to those in Example 1. For specific implementation details, please refer to the relevant description section of Example 1.

[0125] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0126] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A self-disturbance rejection and fault-tolerant control method based on the Scorpion Eagle hybrid algorithm, characterized in that, Includes the following steps: The system acquires the operating parameters and measured force data of the machine to be controlled. Based on the operating parameters and the working principle of the operating device, it calculates the operating force data. The calculated operating force data is subtracted from the measured force data to obtain the difference. The difference is used to construct a fitness function, and the neural network is trained based on the fitness function. By incorporating the firefly algorithm into the guildhound algorithm, a hybrid guildhound algorithm is obtained, in which... The formula for calculating the firefly's position is incorporated into the prey attack formula of the Scorpion Algorithm. A formula for the fitness function of two random candidate solutions of the Scorpion Hybrid Algorithm is constructed to improve the dynamic decision domain radius in the firefly algorithm. The improved dynamic decision domain radius in the firefly algorithm is then incorporated into the escape strategy of the Scorpion Algorithm. The neural network was iteratively optimized using the Scorpion Hybrid Algorithm to obtain a neural network prediction model for the job performance difference. The trained neural network prediction model for the difference in working force is transformed into a multi-segment regression model for the difference in working force. The multi-segment regression model is used to calculate the actual difference in the real-time working parameters collected. The working force calculated in real time based on the working principle of the working device is combined with the actual difference to form an active disturbance rejection and fault tolerance strategy, which is then input into the controller to control the machine to be controlled.

2. The active disturbance rejection and fault-tolerant control method based on the Scorpionhound hybrid algorithm as described in claim 1, characterized in that, Operating parameters include travel speed, operating device speed, operating device upward acceleration, operating device downward acceleration, output torque and speed of the power unit of agricultural machinery or construction machinery, and the rate of change of output torque and speed of the power unit of agricultural machinery or construction machinery. Specifically, based on the operating parameters and the working principle of the operating device, the operating force data is calculated as follows: The data of the working force measured by the force sensor is as follows Data on actual working capacity Data on calculated working force Subtract the two values ​​to obtain the difference, and construct a fitness function based on the difference in work capacity. Among them, f e t (·) represents the objective function value corresponding to the solution.

3. The active disturbance rejection and fault-tolerant control method based on the Scorpion Hybrid Algorithm as described in claim 1, characterized in that, The specific steps for integrating the firefly location calculation formula into the prey attack formula of the Scorpion Algorithm are as follows: First, based on the firefly transfer probability p... kj The optimization process for determining the value of the target parameter involves deciding whether to use the original attack prey formula of the Scorpion Algorithm or the improved attack prey formula. Specifically, when the firefly transfer probability is greater than or equal to a set threshold, the Scorpion Algorithm and the Firefly Algorithm are combined, and ρ is added to the Firefly Algorithm. t After modification, the improved prey attack formula of the Scorpion Algorithm is obtained. When the firefly transfer probability is less than or equal to a set threshold, the original prey attack formula of the Scorpion Algorithm is used. The formula is as follows: In the formula: This represents the current position of the q-th dimension of the necromancer in the first phase, i.e., the prey-attacking phase. This represents the optimal position of the guilder at the t-th iteration, where t represents the current iteration number; T represents the maximum number of iterations, x p,q This represents the current position of the p-th q-th secretary bird, where RL represents the weighted Levy(Dim), Levy(Dim) is the Levy flight parameter, and Dim is the dimension, x... i (t) represents the position of firefly i at time t; x j (t) represents the position of firefly j at time t, and s y This represents the initial step size of the firefly algorithm; During the optimization process, a counter `count` is introduced, with an initial value of 0, and `ρ`. t The calculation formula is: the optimal position of the guilder at the (t-1)th iteration. The optimal position of the guilder at iteration t ratio ρ t The closer the value is to 1, the more similar the results of local optimization; if |ρ t -1|<δ, take δ=0.01, then count=count+1; otherwise count=0; When count accumulates to a certain number of times T, it indicates that the results are very close and the system is trapped in a local optimum. At this point, it is necessary to increase the dynamic policy domain radius.

4. The active disturbance rejection and fault-tolerant control method based on the Scorpion Hybrid Algorithm as described in claim 3, characterized in that, The specific steps for improving the dynamic decision domain radius in the firefly algorithm by constructing a formula for the fitness function of two random candidate solutions from the guilder hybrid algorithm are as follows: The radius in the firefly algorithm can be adjusted by setting an adjustment factor C: In the formula: C is the adjustment coefficient. and It is the position of the random candidate solution in the first stage of selection; f e t (·) represents the objective function value corresponding to the solution; in this formula... This reflects the iterative process; as the number of iterations t gradually approaches the maximum number of iterations T, It will gradually increase from 0 to 1, but C cannot expand indefinitely. When C≥100, let C=100; To measure the degree of difference between solutions in the current search region by the difference in the objective function values ​​of two candidate solutions, if the objective function values ​​of the two candidate solutions are significantly different, it indicates that the diversity of solutions in the region is high and there may be better solutions that have not yet been discovered. Conversely, if the difference is small, it indicates that the solutions in the region are similar and a larger radius needs to be chosen when escaping the local optimum. The formula for updating the radius of the dynamic decision domain is: After introducing the adjustment factor C, the formula becomes In the formula: This represents the radius of the decision domain for firefly k in generation t+1. The dynamic decision domain of firefly k in generation t, N k (t) represents the set of k neighbors of the firefly in generation t, r s Let n be the firefly's perception domain, β be the update rate of the dynamic decision domain, and n be the number of fireflies. t To set a threshold for the neighborhood; As C increases, C×β(n) t -|N k The value of (t)|) will also increase, and at this time, the increase of C will increase the radius of the decision domain. The increase is larger, thus expanding the search range and allowing more local optima to be obtained; r s This represents the sensory domain of a firefly.

5. The active disturbance rejection and fault-tolerant control method based on the Scorpionhound hybrid algorithm as described in claim 4, characterized in that, The formula for incorporating the dynamic decision domain radius from the improved firefly algorithm into the escape strategy of the Scorpion algorithm is as follows: In the formula: express and The ratio, This represents the current position of the q-th dimension of the guildea during the second phase, i.e., the escape phase. Let represent the optimal position of the guilder at the t-th iteration, and RB be a 1×Dim array randomly generated from a standard normal distribution.

6. The active disturbance rejection and fault-tolerant control method based on the Scorpion Hybrid Algorithm as described in claim 5, characterized in that, The Elman neural network was adopted, with the fitness function as the objective. The connection weight matrix of the Elman neural network was iteratively optimized using the Scorpion-Vulture hybrid algorithm to obtain the job performance difference neural network prediction model.

7. A self-disturbance rejection and fault-tolerant control system based on the guilder hybrid algorithm, characterized in that, include: The model training module is configured to acquire the operating parameters and measured force data of the machine to be controlled, calculate the operating force data based on the operating parameters and the working principle of the operating device, subtract the calculated operating force data from the measured force data to obtain the difference, construct a fitness function with the difference as the target, and train the neural network based on the fitness function. The algorithm improvement module is configured to introduce the firefly algorithm into the guilder algorithm to obtain the guilder hybrid algorithm. Specifically, the formula for calculating the firefly's position is incorporated into the prey attack formula of the guilder algorithm. The fitness function formula of the guilder hybrid algorithm is constructed to improve the dynamic decision domain radius in the firefly algorithm. The improved dynamic decision domain radius of the firefly algorithm is incorporated into the guilder algorithm's escape strategy. The guilder hybrid algorithm is used to iteratively optimize the neural network to obtain the job performance difference neural network prediction model. The mechanical control module is configured to transform the trained working force difference neural network prediction model into a multi-segment working force difference regression model. The multi-segment working force difference regression model is used to calculate the actual difference based on the collected real-time working parameters. The working force calculated in real time based on the working principle of the working device is combined with the actual difference to form an active disturbance rejection and fault tolerance strategy, which is then input into the controller to control the machine.

8. A computer-readable storage medium, characterized in that, It stores multiple instructions, which are adapted to be loaded and executed by the processor of the terminal device. The active disturbance rejection and fault-tolerant control method based on the Scorpion Hybrid Algorithm as described in any one of claims 1-6 is described in the following.

9. A terminal device, characterized in that, The invention includes a processor and a computer-readable storage medium, wherein the processor implements various instructions; and the computer-readable storage medium stores multiple instructions adapted to be loaded by the processor and executed by the processor for the active disturbance rejection and fault-tolerant control method based on the guilder hybrid algorithm as described in any one of claims 1-6.

Images