Fault monitoring method for electric vehicle charging equipment

Abstract

The invention discloses a fault monitoring method for electric vehicle charging equipment, and belongs to the technical field of electric vehicles. Comprising the steps of 1, acquiring normal data and historical fault data of charging equipment, and performing data preprocessing; step 2, constructing a data set based on the preprocessed data, and dividing the data set into a training set and a test set; step 3, establishing an electric vehicle charging equipment fault model based on the deep belief network, optimizing the number of hidden layer neurons and reverse learning parameters in the deep belief network through a seagull algorithm, and inputting a multi-feature fusion vector into the model to complete fault identification; and 4, acquiring real-time state data of the charging equipment, preprocessing the real-time state data, and inputting the processed data into the charging equipment fault diagnosis model to obtain a fault monitoring result of the charging equipment. Parameter optimization is carried out on the deep belief network in the monitoring process through a sea gull algorithm, and fault monitoring of the charging equipment in a complex environment is realized.

Description

Technical Field

[0001] The invention relates to a fault monitoring method for electric vehicle charging equipment, belonging to the technical field of electric vehicles. Background Art

[0002] With the massive increase in the number of electric vehicles and charging infrastructure, the operating status, monitoring and fault diagnosis of charging equipment have become one of the main issues affecting the safe and stable operation of charging equipment. Since charging equipment should be convenient for charging electric vehicles, most of the installation locations are outdoors, which leads to long-term exposure of charging equipment to harsh environments such as rain and dew. At the same time, if the core components of the charging equipment such as the charging module and energy conversion module fail, it will inevitably affect the safety and reliability of the entire charging equipment system. Therefore, it is necessary to develop a practical method for monitoring faults in electric vehicle charging equipment, which is of great significance for maintaining the normal operation of charging equipment and reducing maintenance costs. Summary of the invention

[0003] The present invention designs and develops a fault monitoring method for electric vehicle charging equipment. The Seagull algorithm is used to optimize the parameters of the deep belief network during the monitoring process, thereby realizing fault monitoring of electric vehicle charging equipment in a complex environment and improving the detection accuracy.

[0004] The technical solution provided by the present invention is: A fault monitoring method for an electric vehicle charging device, comprising: Step 1: Obtain normal data and historical fault data of the charging equipment and perform data preprocessing; Step 2: construct a data set based on the preprocessed data, and divide the data set into a training set and a test set; Step 3: Establish an electric vehicle charging equipment fault model based on a deep belief network, optimize the number of hidden layer neurons and reverse learning parameters in the deep belief network through the Seagull algorithm, input the multi-feature fusion vector into the model, and complete fault identification; Step 4: Acquire the real-time status data of the charging device, perform preprocessing on it, input the processed data into the charging device fault diagnosis model, and obtain the fault monitoring result of the charging device.

[0005] Preferably, in step 2, 75% of the data in the data set is used as a training set, and the remaining 25% of the data is used as a test set.

[0006] Preferably, the deep belief network includes two RBMs and one BP classifier.

[0007] Preferably, the Seagull algorithm optimization process includes: Step 1: Initialize the parameters of the Seagull algorithm and the deep belief network, and use the training error as the fitness function; Step 2: Initialize the position of the seagull, calculate the individual fitness value, compare the individual fitness values, and calculate the best position of the seagull; Step 3: Calculate the optimal fitness value of the prey, update the attack position of the seagull, and obtain the optimal parameters of the deep belief network.

[0008] Preferably, in the Seagull algorithm, the training set prediction error is selected as the fitness function value, and the calculation formula is: ; In the formula, is the fitness value, is the number of correctly classified samples, is the total sample size.

[0009] Preferably, it also includes: Using parameters The formula for coordinating the positions of the seagulls and determining a new position that will not collide with other seagulls is: ; ; In the formula, is the current location of the seagull, is the current iteration number, is a constant, with a value of 2. is the maximum number of iterations.

[0010] Preferably, the calculating the best position of the seagull comprises: Determine the best moving direction of the seagull, the calculation formula is: ; In the formula, is the current best seagull position, It is used to coordinate the global and local search of the algorithm, and its calculation formula is; ; In the formula, for A random number between After the seagull finds the best moving direction, it moves to the best position and reaches the new position. The calculation formula for the seagull's new position is: .

[0011] Preferably, when the seagull attacks, , , The calculation formula for motion behavior in a plane is: ; In the formula, is the spiral flight radius of the seagull, for Random angles between and is a constant, with a value of 1. is the base of natural logarithms; The update equation for the attack position is: .

[0012] Preferably, the step three further comprises: Reverse learning is based on comparing the obtained reverse individuals with the current group to find the best result; set up The individuals with the top 50% fitness exist The value of the dimension, the calculation formula of the reverse individual is: ; In the formula, for A random number between , The interval consisting of the first 50% of the population; After the seagull attack position is updated, sort by fitness and set is the fitness function after the attack, taking the top 50% of the seagull fitness values ​​and using the reverse learning strategy to solve the reverse individuals; Compare the best individual before and after the reverse, if it is better, replace the previous best seagull, the update method is: ; In the formula, For a better individual seagull position, is the fitness function of the better individual position of seagulls, is the fitness function of the best individual seagull position.

[0013] The beneficial effects described in the present invention: The fault monitoring method for electric vehicle charging equipment provided by the present invention optimizes the parameters of the deep belief network during the monitoring process through the Seagull algorithm, thereby realizing fault monitoring of electric vehicle charging equipment in a complex environment and improving the detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a structural schematic diagram of the fault monitoring method of the electric vehicle charging equipment according to the present invention.

[0015] Figure 2 It is a comparison chart of test results in the embodiments described in the present invention. DETAILED DESCRIPTION

[0016] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0017] like Figure 1 As shown, the present invention provides a fault monitoring method for electric vehicle charging equipment, which optimizes the parameters of the deep belief network during the monitoring process through the Seagull algorithm to realize the fault monitoring of the electric vehicle charging equipment in a complex environment, including: Step 1: Obtain normal data and historical fault data of the charging equipment and perform data preprocessing; After cleaning and sorting the historical status data of electric vehicle charging equipment, the data is classified and the corresponding operating status label is added to each type of data to construct a data set; Step 2: Construct a data set based on the preprocessed data and divide the data set into a training set and a test set; select 75% of the data in the data set as the training set and the remaining 25% of the data as the test set; Step 3: Establish an electric vehicle charging equipment fault model based on a deep belief network, optimize the number of hidden layer neurons and reverse learning parameters in the deep belief network through the Seagull algorithm, input the multi-feature fusion vector into the model, and complete fault identification; Among them, the deep belief network includes two RBMs and a BP classifier; Step 4: Acquire the real-time status data of the charging device, perform preprocessing on it, input the processed data into the charging device fault diagnosis model, and obtain the fault monitoring result of the charging device.

[0018] Among them, the optimization process of the Seagull algorithm includes: Step 1: Initialize the parameters of the Seagull algorithm and the deep belief network, and use the training error as the fitness function; Step 2: Initialize the position of the seagull, calculate the individual fitness value, compare the individual fitness values, and calculate the best position of the seagull; Step 3: Calculate the best fitness value of the prey, update the attack position of the seagull, update the position and fitness value of the seagull, take the top 50% of the fitness to form a new group, and obtain the boundary; Find the reverse individuals of the top 50% of individuals in fitness value, compare the fitness values ​​before and after the reverse, and update the best seagull position and fitness value; Determine whether the maximum number of iterations has been reached. If so, proceed to the next step. Otherwise, repeat step 3 until the global optimal initial weights and thresholds are output and the optimal parameters of the deep belief network are obtained.

[0019] The Seagull algorithm optimizes the number of neurons in each hidden layer of the deep belief network and reversely fine-tunes the learning rate to eliminate the influence of relying on empirical formulas on the diagnosis results. The prediction error of the training set is selected as the fitness function. ; In the formula, is the fitness value, is the number of correctly classified samples, is the total sample size.

[0020] Using parameters Coordinate the positions of the seagulls to prevent collisions between them. The calculation formula for the new position that will not collide with other seagulls is: ; ; In the formula, is the current location of the seagull, is the current iteration number, is a constant, with a value of 2. is the maximum number of iterations.

[0021] Calculate the best position and direction for the seagull, including: When there is no conflict with other seagulls, the seagull will move to the best position. The best moving direction of the seagull is determined by the following calculation formula: ; In the formula, is the current best seagull position, It is used to coordinate the global and local search of the algorithm, and its calculation formula is: ; In the formula, for A random number between .

[0022] After the seagull finds the best moving direction, it moves to the best position while avoiding collisions with other seagulls and reaches a new position. The calculation formula for the new position of the seagull is: .

[0023] Seagulls fly in spirals at different angles and speeds in the air to hunt and attack. , , The calculation formula for motion behavior in a plane is: ; In the formula, is the spiral flight radius of the seagull, for Random angles between and is a constant, with a value of 1. is the base of natural logarithms.

[0024] The update equation for the attack position is: ; The reverse learning strategy can improve the problem of poor population diversity in the later stage of the algorithm and effectively improve the optimization accuracy by comparing the obtained reverse individuals with the current group.

[0025] Reverse learning is based on comparing the obtained reverse individuals with the current group to find the best result; set up The top 50% individuals in terms of fitness exist The value of the dimension, the calculation formula of the reverse individual is: ; In the formula, for A random number between , The interval consisting of the first 50% of the population; After the seagull attack position is updated, sort by fitness and set is the fitness function after the attack. The top 50% of the seagull fitness values ​​are taken and the reverse individual is solved using the reverse learning strategy. The best individual before and after the reverse is compared. If it is better, the previous best seagull is replaced. The update method is: ; In the formula, For a better individual seagull position, is the fitness function of the better individual position of seagulls, Fitness function of the best individual seagull position. Example

[0026] Based on the fault monitoring method of electric vehicle charging equipment, under the operating conditions of Windows 10 system, the MATLAB / Simulink platform is used to intervene in the current test program, combined with the PWM distillation power device for measurement. In the selected area, 30 electric vehicle charging piles are selected as test targets, randomly divided into 5 groups, and 3 groups of virtual abnormal fault instructions are specified and entered into the current test structure to realize the setting of the simulation environment, and set the auxiliary test indicators and values, as shown in the following table: Table 1 Auxiliary test indicators and value setting table Auxiliary test indicator name Numerical reference Grid voltage effective value / V 220 AC side inductance / mH 6.5 AC side resistance / Ω 0.45 Grid voltage frequency 45-65 DC side voltage setting value / V 650 A certain number of identification and prediction nodes are deployed near the charging piles of electric vehicles, and the nodes are connected to each other to form a monitoring closed loop. Fault detection is carried out by signal recognition, and deep learning technology is combined to build a multi-point, multi-time point multi-fault prediction program, which is connected with the main control program to complete the construction of the test environment.

[0027] In the test environment that has been built, the automatic prediction method of electric vehicle charging pile faults is practiced and verified by combining deep learning technology. The preset virtual simulation fault instruction group is imported into the current test environment. After execution, the program will immediately identify and calibrate the abnormal charging pile of the electric vehicle, and automatically predict the location where the fault may occur, and collect the corresponding prediction signal. Irregular signal change points will appear at the predicted abnormal location, forming a certain abnormal signal band. After extracting the fault characteristics from the band and forming the characteristic value, it is necessary to predict the fault frequency first. The calculation formula for predicting the fault frequency is: ; In the formula, To predict the failure frequency, is the fault prediction frequency, , are the initial fault area and the overlapping fault area respectively. Repeat the fault location for the charging pile, It takes time to predict failure.

[0028] Combined with the current measurement, the predicted fault frequency is calculated. According to the change of fault frequency, the fault level of the charging pile is determined. At the same time, combined with the obtained fault automatic prediction signal, the predicted number of fault intervals is calculated. The calculation formula for the average interval prediction number is: ; In the formula, is the mean number of predicted failure intervals, is the total fault coverage, is the fault edge point, is the fault extendable area, Combined with the current measurement, the test results are analyzed. Figure 2 As shown in the figure, for the selected 5 charging pile groups, 3 groups of virtual fault instructions are imported in sequence, and the final average fault interval can reach more than 15 times, indicating that the deep learning electric vehicle charging pile automatic prediction method designed this time is more flexible and efficient, and can also automatically predict and identify faults of electric vehicle charging equipment in complex environments.

[0029] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A fault monitoring method for electric vehicle charging equipment, characterized in that: include: Step 1: Obtain normal data and historical fault data of the charging equipment and perform data preprocessing; Step 2: construct a data set based on the preprocessed data, and divide the data set into a training set and a test set; Step 3: Establish an electric vehicle charging equipment fault model based on a deep belief network, optimize the number of hidden layer neurons and reverse learning parameters in the deep belief network through the Seagull algorithm, input the multi-feature fusion vector into the model, and complete fault identification; Step 4: Acquire the real-time status data of the charging device, preprocess it, input the processed data into the charging device fault diagnosis model, and obtain the fault monitoring result of the charging device.

2. The fault monitoring method for electric vehicle charging equipment according to claim 1, characterized in that: In step 2, 75% of the data in the data set is used as a training set, and the remaining 25% of the data is used as a test set.

3. The fault monitoring method for electric vehicle charging equipment according to claim 2, characterized in that: The deep belief network includes two RBMs and a BP classifier.

4. The fault monitoring method for electric vehicle charging equipment according to claim 3, characterized in that: The Seagull algorithm optimization process includes: Step 1: Initialize the parameters of the Seagull algorithm and the deep belief network, and use the training error as the fitness function; Step 2: Initialize the position of the seagull, calculate the individual fitness value, compare the individual fitness values, and calculate the best position of the seagull; Step 3: Calculate the optimal fitness value of the prey, update the attack position of the seagull, and obtain the optimal parameters of the deep belief network.

5. The fault monitoring method for electric vehicle charging equipment according to claim 4, characterized in that: The Seagull algorithm selects the training set prediction error as the fitness function value, and the calculation formula is: ； In the formula, is the fitness value, is the number of correctly classified samples, is the total sample size.

6. The fault monitoring method for electric vehicle charging equipment according to claim 5, characterized in that: Also includes: Using parameters The formula for coordinating the positions of the seagulls and determining a new position that will not collide with other seagulls is: ； ； In the formula, is the current location of the seagull, is the current iteration number, Constant, value is 2, is the maximum number of iterations.

7. The fault monitoring method for electric vehicle charging equipment according to claim 6, characterized in that: The calculation of the best position of the seagull includes: Determine the best moving direction of the seagull, the calculation formula is: ； In the formula, is the current best seagull position, It is used to coordinate the global and local search of the algorithm, and its calculation formula is; ； In the formula, for A random number between After the seagull finds the best moving direction, it moves to the best position and reaches the new position. The calculation formula for the seagull's new position is: 。 8. The fault monitoring method for electric vehicle charging equipment according to claim 7, characterized in that: The seagulls attack when , , The calculation formula for motion behavior in a plane is: ； In the formula, is the spiral flight radius of the seagull, for Random angles between and is a constant, with a value of 1. is the base of natural logarithms; The update equation for the attack position is: 。 9. The fault monitoring method for electric vehicle charging equipment according to claim 8, characterized in that: The step three also includes: Reverse learning is based on comparing the obtained reverse individuals with the current group to find the best result; set up The individuals with the top 50% fitness exist The value of the dimension, the calculation formula of the reverse individual is: ； In the formula, for A random number between , The interval consisting of the first 50% of the population; After the seagull attack position is updated, sort by fitness and set is the fitness function after the attack, taking the top 50% of the seagull fitness values ​​and using the reverse learning strategy to solve the reverse individuals; Compare the best individuals before and after the reverse, if it is better, replace the previous best seagull, the update method is: ； In the formula, For a better individual seagull position, is the fitness function of the better individual position of seagulls, is the fitness function of the best individual seagull position.

Images