Fault diagnosis method for electro-hydraulic actuating mechanism

Abstract

The invention discloses an electro-hydraulic actuating mechanism fault diagnosis method, and relates to the field of electro-hydraulic actuating mechanism fault diagnosis. The invention aims to solve the problem that the traditional parameter optimization method is easy to fall into local optimum and high in calculation cost due to high dependence on selection of smooth factors in the existing fault diagnosis method for the electro-hydraulic actuating mechanism based on the probabilistic neural network. According to the fault diagnosis method for the electro-hydraulic actuating mechanism, the fault feature data of the electro-hydraulic actuating mechanism are collected and input into the probabilistic neural network fault diagnosis model, and a fault recognition result is obtained. The probabilistic neural network is optimized through an index-trigonometric function optimization algorithm, the optimal smoothing factor can be obtained, and the smoothing factor selection problem can be effectively solved. The method has the advantages of being easy to train, rapid in convergence, high in accuracy and the like, and faults can be rapidly and effectively recognized.

Description

Technical Field

[0001] The present invention belongs to the field of fault diagnosis of electro-hydraulic actuators. Background Art

[0002] Electro-hydraulic actuators are industrial control devices commonly used in fields such as ships and petrochemicals. Fault diagnosis is very important for their reliable operation. Traditional fault diagnosis methods rely on manual experience and have problems such as low diagnosis efficiency and poor generalization ability. Probabilistic neural networks (PNNs) are widely used in fault diagnosis due to their strong non-linear classification ability. However, their performance highly depends on the selection of the smoothing factor, and traditional parameter optimization methods are prone to falling into local optima and have high computational costs. Summary of the Invention

[0003] The present invention aims to solve the problem that in the existing electro-hydraulic actuator fault diagnosis method based on probabilistic neural networks, due to the high dependence on the selection of the smoothing factor, traditional parameter optimization methods are prone to falling into local optima and have high computational costs. Now, a fault diagnosis method for electro-hydraulic actuators based on an exponential-trigonometric optimization algorithm is provided. As a new type of intelligent algorithm, the exponential-trigonometric optimization algorithm (ETO) has advantages such as fast convergence speed and strong global search ability. Combining the exponential-trigonometric algorithm with PNN can solve the problem of PNN parameter optimization, obtain the optimal smoothing factor, and solve the problem of smoothing factor selection.

[0004] A fault diagnosis method for electro-hydraulic actuators includes:

[0005] Collecting fault feature data of the electro-hydraulic actuator and inputting the fault feature data into a probabilistic neural network fault diagnosis model to obtain a fault recognition result.

[0006] Further, the above-mentioned probabilistic neural network fault diagnosis model includes: an input layer, a pattern layer, a summation layer, and an output layer;

[0007] The input layer is used to receive the fault feature data and transfer it to the pattern layer;

[0008] The pattern layer is used to perform Gaussian function calculation on the fault feature data to obtain the neuron category to which the fault feature data belongs;

[0009] The summation layer is used to take the average of the outputs of neurons of the same category to obtain the estimated probability of each category of neurons;

[0010] The output layer is used to output the fault category with the highest estimated probability as the fault recognition result.

[0011] Further, the above-mentioned Gaussian function expression is:

[0012]

[0013] Among them, σ is the smoothing factor, p is the dimension of the fault feature data, and x αβ is the α-th pattern layer neuron of the β-th class, and Φ ij (·) represents the Gaussian function.

[0014] Furthermore, the training process of the above probability neural network fault diagnosis model includes:

[0015] Collect the fault feature data and fault categories of the electro-hydraulic actuator;

[0016] Optimize the smoothing factor of the probability neural network fault diagnosis model based on the exponential-trigonometric function optimization algorithm to obtain the trained probability neural network fault diagnosis model.

[0017] Furthermore, the optimization of the smoothing factor of the probability neural network fault diagnosis model based on the exponential-trigonometric function optimization algorithm includes:

[0018] Traverse each variable individual and calculate the optimal candidate position of each variable individual;

[0019] In the population, take the optimal candidate position with the smallest fitness value as the optimal smoothing factor of the probability neural network fault diagnosis model, so as to obtain the probability neural network fault diagnosis model optimized by the exponential-trigonometric function.

[0020] Furthermore, the calculation of the optimal candidate position of each variable individual includes:

[0021] Initialization: Randomly initialize the population position;

[0022] S1: Calculate the behavior judgment parameter CM at the (t + 1)-th iteration through the following formula:

[0023]

[0024]

[0025]

[0026] When it satisfies , the behavior at the (t + 1)-th iteration is in the second stage, otherwise it is in the first stage, where floor(·) represents the rounding function; S2: In the exploration first stage, update the individual position according to the following formula:

[0027]

[0028] Among them, X best is the optimal candidate position, and respectively represent the positions of the j-th dimensional variable individuals in the i-th population at the iteration times t and t + 1;

[0029] In the second exploration stage, update the individual position according to the following formula:

[0030]

[0031] Among them,

[0032] In the first exploitation stage, update the individual position according to the following formula:

[0033]

[0034] Among them,

[0035] In the second exploitation stage, update the individual position according to the following formula:

[0036]

[0037] Among them,

[0038] S3: Take the update result as the smoothing factor of the PNN model, input the fault feature data into the PNN model, obtain the fault type recognition result, and then calculate the fitness value F corresponding to the update result ;

[0039] S4: Judge whether the fitness value F is less than the minimum fitness value F min . If so, take the fitness value F as the minimum fitness value F min , and take the update result as the optimal candidate position X best . Otherwise, do not update F min and the optimal candidate position;

[0040] S5: Judge whether t + 2 > T holds. If so, take the optimal candidate position X obtained in S4 best as the optimal candidate position of the j-th dimensional variable individuals in the i-th population. Otherwise, set t = t + 1 and return to S1.

[0041] Furthermore, the expression of the above fitness value F is:

[0042]

[0043] Among them, M is the number of fault feature data, and Z is the number of accurate identifications.

[0044] Further, initialize the initial position of each variable individual according to the following formula:

[0045]

[0046] Among them, Up j and Low j are respectively the upper bound and the lower bound of the j-th dimensional variable, i = 1, 2,..., N, j = 1, 2,..., D, N is the population size, D is the variable dimension, is the initial position of the j-th dimensional variable individual in the i-th population.

[0047] Further, after collecting the fault feature data and fault categories of the electro-hydraulic actuator, perform standardization processing on the fault feature data, and then perform smoothing factor optimization.

[0048] Further, the above-mentioned standardization processing of the fault feature data includes:

[0049] Perform standardization processing using the Min-Max standardization method:

[0050]

[0051] Among them, x s is the standardized fault feature data, x is the fault feature data before standardization, x min is the minimum value of the fault feature data, x max is the maximum value of the fault feature data.

[0052] The advantages of the present invention are as follows:

[0053] 1. The present invention uses the Min-Max standardization method to process the electro-hydraulic actuator data, which can not only retain the distribution and relative relationship of the original data, but also ensure the scale consistency of different features, and can effectively ensure the reliability of the diagnostic data.

[0054] 2. The present invention optimizes the probabilistic neural network through the exponential-trigonometric function optimization algorithm, and can obtain the optimal smoothing factor, which can effectively solve the problem of selecting the smoothing factor.

[0055] 3. The present invention has the advantages of easy training, fast convergence, high accuracy, etc., and can quickly and effectively identify faults. Brief Description of the Drawings

[0056] Figure 1 is a flowchart of a fault diagnosis method for an electro-hydraulic actuator according to the present invention;

[0057] Figure 2 It is a structural diagram of a probabilistic neural network. Specific implementation mode

[0058] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other.

[0059] Refer to Figure 1 and Figure 2 Specifically illustrate this implementation mode. A fault diagnosis method for an electro-hydraulic actuator described in this implementation mode includes the following steps:

[0060] Step 1: Collect electro-hydraulic actuator data to obtain fault feature data and fault categories.

[0061] The described fault feature data includes: fault feature data 1, fault feature data 2,..., fault feature data m.

[0062] The described fault categories include: fault category 1, fault category 2,..., fault category n.

[0063] Step 2: Process the fault feature data using a standardization method to obtain the fault feature data.

[0064] The standardization method is the Min-Max standardization method, and the formula is as follows:

[0065]

[0066] where x s is the standardized fault feature data, x is the fault feature data before standardization, x min is the minimum value of the fault feature data, and x max is the maximum value of the fault feature data.

[0067] Step 3: Divide X% of the fault feature data into training data and Y% of the fault feature data into test data, where X + Y = 100.

[0068] Step 4: Construct a probabilistic neural network fault diagnosis model, use the fault feature data as the input of the probabilistic neural network fault diagnosis model, and use the fault category as the output of the probabilistic neural network fault diagnosis model.

[0069] The probabilistic neural network fault diagnosis model includes: an input layer, a pattern layer, a summation layer, and an output layer.

[0070] The input layer is used to receive the fault feature data and transfer it to the pattern layer.

[0071] The pattern layer is used to perform Gaussian function calculation on the fault feature data of the input layer to obtain the output of the pattern layer. The formula is as follows:

[0072]

[0073] where σ is the smoothing factor, p is the dimension of the fault feature data, x αβ is the α-th pattern layer neuron of the β-th class, and Φ ij (·) represents the Gaussian function.

[0074] The summation layer is used to take the average of the outputs of the neurons of the same class to obtain the estimated probabilities of the neurons of each class.

[0075] The output layer is used to output the fault class with the highest estimated probability as the fault recognition result.

[0076] Step 5: Optimize the probability neural network fault diagnosis model with the exponential-trigonometric function optimization algorithm and use the training data to train to obtain the optimal smoothing factor. The specific steps are as follows:

[0077] Initialize the algorithm parameters: population size N, variable dimension D, maximum number of iterations T, t = 0, 1, 2,..., T;

[0078] Randomly initialize the population: Randomly generate the initial population, and the initial position of each variable individual is determined by the following formula:

[0079]

[0080] where rand() is a random value in [0, 1], Up j and Low j are the upper and lower bounds of the j-th variable respectively, i = 1, 2,..., N, j = 1, 2,..., D, is the initial position of the j-th variable in the i-th population.

[0081] Traverse each variable individual and calculate the optimal candidate position of each variable individual. In the population, the optimal candidate position with the smallest fitness value is used as the optimal position, and this optimal position is the optimal smoothing factor of the probability neural network fault diagnosis model, so as to obtain the probability neural network fault diagnosis model optimized by the exponential-trigonometric function.

[0082] The steps of calculating the optimal candidate position corresponding to each variable individual include:

[0083] S1: Determine whether the behavior in the (t + 1)-th iteration is in the exploration stage or the exploitation stage through the following formula:

[0084]

[0085] Among them, CM is the behavior judgment parameter. When CM > 1, the behavior in the (t + 1)-th iteration is in the exploration stage; when CM < 1, the behavior in the (t + 1)-th iteration is in the exploitation stage.

[0086]

[0087] When is satisfied, the behavior in the (t + 1)-th iteration is in the second stage; otherwise, it is in the first stage, where floor(·) represents the rounding function.

[0088] S2: Update the individual position

[0089] a). In the first exploration stage, update the individual position according to the following formula:

[0090]

[0091] where X best is the optimal candidate position, and respectively represent the positions of the j-th variable in the i-th population at the iteration times t and t + 1.

[0092] b). In the second exploration stage, update the individual position according to the following formula:

[0093]

[0094] c). In the first exploitation stage, update the individual position according to the following formula:

[0095]

[0096] d). In the second exploitation stage, update the individual position according to the following formula:

[0097]

[0098] S3: Use the updated result as the smoothing factor of the PNN model, input the fault feature data into the PNN model to obtain the fault type recognition result, and then calculate the fitness value F corresponding to the updated result :

[0099]

[0100] where M is the number of fault feature data and Z is the number of accurately recognized ones.

[0101] S4: Judge whether the fitness value F is less than the minimum fitness value Fmin , if so, take the fitness value F as the minimum fitness value F min , and take the update result as the optimal candidate position X best , otherwise do not update F min and the optimal candidate position.

[0102] S5: Determine whether t + 2 > T holds. If so, take the optimal candidate position X obtained in S4 best as the optimal candidate position of the j - dimensional variable individual in the i - th population. Otherwise, set t = t + 1 and return to S1.

[0103] Step 6, input the test data into the probability neural network fault diagnosis model optimized by the exponential - triangle optimization algorithm, and output the fault diagnosis result. Specific embodiments

[0105] Now, conduct fault diagnosis on a certain electro - hydraulic actuator. Figure 1 This is the fault diagnosis flowchart of this embodiment. The electro - hydraulic actuator data includes hydraulic oil pressure, hydraulic oil temperature, hydraulic oil flow rate, stroke position, and motor current. The fault categories include leakage, spool jamming, slow response, and motor failure.

[0106] Divide 70% of the fault feature data into training data and 30% of the fault feature data into test data.

[0107] Initialization: The population size N = 20, the variable dimension D = 1, and the maximum number of iterations T = 50.

[0108] Test the method of the present invention with the electro - hydraulic actuator data, and the obtained fault diagnosis accuracy rate is as high as 97.86%.

[0109] In summary, the present invention discloses a fault diagnosis method for an electro - hydraulic actuator aiming at the problem of electro - hydraulic actuator fault diagnosis. The main features of the present invention are as follows: collect electro - hydraulic actuator data to obtain fault feature data and fault categories; process the fault feature data with a standardization method to obtain the fault feature data; divide the fault feature data into training data and test data; construct a probability neural network fault diagnosis model, take the fault feature data as the input of the probability neural network fault diagnosis model, and take the fault categories as the output of the probability neural network fault diagnosis model; optimize the probability neural network fault diagnosis model with the exponential - triangle optimization algorithm, and train with the training data to obtain the optimal smoothing factor; input the test data into the probability neural network fault diagnosis model optimized by the exponential - triangle optimization algorithm, and output the fault diagnosis result. The present invention has the advantages of easy training, fast convergence, high accuracy rate, etc., and can quickly and effectively identify faults.

[0110] Although the present invention has been described herein with reference to particular embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. Accordingly, it should be understood that numerous modifications may be made to the exemplary embodiments, and other arrangements may be devised, without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the different dependent claims and the features described herein may be combined in ways different from those described in the original claims. It should also be understood that the features described in connection with separate embodiments may be used in other described embodiments.

Claims

1. A method for diagnosing faults of an electro-hydraulic actuator, characterized in that: include: The fault characteristic data of the electro-hydraulic actuator is collected and input into the probabilistic neural network fault diagnosis model to obtain the fault identification result.

2. A fault diagnosis method for an electro-hydraulic actuator according to claim 1, characterized in that: The probabilistic neural network fault diagnosis model includes: an input layer, a pattern layer, a summation layer and an output layer; The input layer is used to receive the fault characteristic data and pass it to the mode layer; The pattern layer is used to perform Gaussian function calculation on the fault feature data to obtain the neuron category to which the fault feature data belongs; The summation layer is used to average the outputs of neurons of the same category to obtain the estimated probability of neurons of each category; The output layer is used to output the fault category with the highest estimated probability as the fault identification result.

3. A fault diagnosis method for an electro-hydraulic actuator according to claim 2, characterized in that: The Gaussian function expression is: Among them, σ is the smoothing factor, p is the dimension of fault feature data, x αβ is the αth pattern layer neuron of the βth class, Φ ij (·) represents a Gaussian function.

4. The electro-hydraulic actuator fault diagnosis method according to claim 1, characterized in that: The training process of the probabilistic neural network fault diagnosis model includes: Collect fault characteristic data and fault categories of electro-hydraulic actuators; The smoothing factor of the probabilistic neural network fault diagnosis model is optimized based on the exponential-trigonometric function optimization algorithm to obtain a trained probabilistic neural network fault diagnosis model.

5. The electro-hydraulic actuator fault diagnosis method according to claim 4, characterized in that: The method of optimizing the smoothing factor of the probabilistic neural network fault diagnosis model based on the exponential-trigonometric function optimization algorithm includes: Traverse each variable individual and calculate the optimal candidate position of each variable individual; In the population, the optimal candidate position with the smallest fitness value is taken as the optimal smoothing factor of the probabilistic neural network fault diagnosis model, so as to obtain the probabilistic neural network fault diagnosis model after exponential-trigonometric function optimization.

6. A method for diagnosing faults of an electro-hydraulic actuator according to claim 5, characterized in that: The step of calculating the optimal candidate position of each variable individual includes: Initialization: Randomly initialize the population position; S1: Calculate the behavior judgment parameter CM at t+1 iterations by the following formula: Wherein, CM is the behavior judgment parameter. When CM>1, the behavior at t+1 iteration is in the exploration stage. When CM<1, the behavior at t+1 iteration is in the development stage. rand() is a random value in [0,1]. T is the maximum number of iterations, t=0,1,2,...,T. When satisfied When , the behavior at the t+1th iteration is in the second stage, otherwise it is in the first stage, where floor(·) represents the rounding function; S2: In the first stage of exploration, the individual position is updated according to the following formula: in, X best is the optimal candidate position, and They represent the position of the j-th dimension variable individual in the i-th population at iterations t and t+1 respectively; In the second phase of exploration, the individual position is updated according to the following formula: in, In the first stage of development, the individual positions are updated according to the following formula: in, In the second stage of development, the individual positions are updated according to the following formula: in, S3: will update the results As the smoothing factor of the PNN model, the fault feature data is input into the PNN model to obtain the fault type identification result, and then the update result is calculated. The corresponding fitness value F; S4: Determine whether the fitness value F is less than the minimum fitness value F min , then the fitness value F is taken as the minimum fitness value F min , and will update the results As the optimal candidate position X best Otherwise, F min Update with the optimal candidate position; S5: Determine whether t+2>T is true, if so, change the optimal candidate position X obtained in S4 to best As the optimal candidate position of the j-th dimension variable individual in the ith population, otherwise set t=t+1 and return S1.

7. A method for diagnosing faults of an electro-hydraulic actuator according to claim 6, characterized in that: The expression of the fitness value F is: Among them, M is the number of fault feature data, and Z is the number of accurate identifications.

8. The electro-hydraulic actuator fault diagnosis method according to claim 6, characterized in that: Initialize the initial position of each variable individual according to the following formula: Among them, Up j and Low j are the upper and lower bounds of the j-th dimension variable, i = 1, 2, ..., N, j = 1, 2, ..., D, N is the population size, D is the variable dimension, is the initial position of the individual with the j-th dimension variable in the ith population.

9. The electro-hydraulic actuator fault diagnosis method according to claim 4, characterized in that: After the fault characteristic data and fault categories of the electro-hydraulic actuator are collected, the fault characteristic data are standardized, and then the smoothing factor is optimized.

10. The electro-hydraulic actuator fault diagnosis method according to claim 9, characterized in that: The standardization process of the fault characteristic data includes: The Min-Max standardization method is used for standardization: Among them, x s is the fault characteristic data after standardization, x is the fault characteristic data before standardization, and x min is the minimum value of the fault characteristic data, x max is the maximum value of the fault characteristic data.

Images