Motor main insulation state early fault on-line monitoring method based on improved SVR
Through the improved SVR method and Gray Wolf algorithm, combined with winding temperature, early online fault monitoring of the main insulation state of three-phase asynchronous motors is achieved, solving the problem that the motor health status cannot be monitored in real time in the existing technology, and improving the monitoring accuracy and robustness reduces equipment downtime and ensures production safety.
Patent Information
- Application Number
- CN202411206821.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-11-19
AI Technical Summary
The existing online monitoring method for main insulation of motors cannot achieve early fault diagnosis, and is poorly robust, so it cannot monitor the health status in real time during the motor operation, resulting in potential safety hazards.
Using an improved support vector regression (SVR) method combined with Gray Wolf algorithm and adaptive weight strategy, the early online monitoring method for the main insulation state of three-phase asynchronous motors based on improved SVR is constructed in real time. Monitor the health status of insulation.
It realizes early online monitoring of the main insulation state of the motor, improves the accuracy and robustness of the monitoring, can promptly detect potential problems, reduce equipment downtime, and ensure production safety.
Smart Images

Figure CN118980930A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an online monitoring method for early fault of main insulation state of a three-phase asynchronous motor based on improved SVR, and belongs to the field of industrial process control. Background Art
[0002] Motors are the most typical energy conversion equipment and are widely used in industries such as petroleum, chemical, and papermaking. Stator insulation failure is a typical failure mode. Statistics show that more than 30% of motor failures are related to stator insulation failure. Compared with mechanical failures (such as bearing failures), electrical failures often occur more suddenly, resulting in serious consequences. Therefore, early insulation fault diagnosis is of great significance for avoiding accidents and improving economic benefits.
[0003] The main insulation (ground wall insulation) of the motor is the most critical component in the motor insulation system. Typical fault diagnosis methods include insulation resistance test, polarization index test, dissipation factor test, etc. However, these tests can only be completed offline during the motor maintenance phase, and cannot be completed during the operation of the motor. Considering that the maintenance cycle interval of the motor is long (such as several months, 1 year or longer, depending on the specific application), it is impossible to complete the diagnosis in time at the early stage of the motor fault. It is more practical to monitor the health status of the motor online during the operation of the motor to complete the early fault diagnosis. Most of the existing online monitoring methods for the main insulation of the motor are based on temperature warning and other methods. However, at this time, most of the main insulation is already in the late degradation stage, and there is not much time left before complete failure. For example, statistics show that the time interval from the occurrence of inter-turn short-circuit insulation fault to the complete failure of the winding ranges from a few seconds to a few hours. Therefore, the existing main insulation online monitoring methods have limited contribution to early fault diagnosis and predictive maintenance. Some scholars have proposed an online monitoring method for main insulation based on equivalent capacitance. The basic idea is to calculate the common mode impedance based on the three voltages and leakage current of the AC motor, and then calculate the equivalent capacitance of the winding main insulation system. The results of the accelerated test show that equivalent capacitance can reflect the state of insulation degradation. However, equivalent capacitance is not a recognized indicator for evaluating the health status of insulation, and its correlation with the actual health status of insulation is still unclear. In addition, since the equivalent capacitance is calculated by measuring electrical parameters such as three voltages, and the measured values of the electrical parameters are related to the measured temperature, in the actual engineering application process, considering that the operation and environmental conditions of the equipment to be monitored are usually in a continuous change process, the temperature of the winding changes accordingly. Therefore, the main insulation online monitoring method based on equivalent capacitance has poor robustness. Considering that insulation resistance is an effective indicator for judging insulation performance, there are corresponding indicators for reference in international and national standards. Therefore, the present invention proposes a method for mapping equivalent capacitance to insulation resistance, thereby improving the accuracy of insulation health monitoring. Considering that there is a complex nonlinear relationship between insulation resistance and the phase voltage and leakage current used to determine the equivalent capacitance, an artificial intelligence method is introduced in the modeling process. Considering that the number of fault samples is small in actual engineering applications, it is proposed to model based on the SVR method. At the same time, in order to improve the accuracy, it is proposed to optimize the GWO algorithm by using the improved Logistic chaotic mapping algorithm and the adaptive weight strategy, and combine it with SVR. In addition, in order to improve the robustness of the method, the winding temperature is introduced into the model input parameter, so that the model can adapt to complex and changeable working conditions, and can still complete accurate monitoring even when the measured temperature changes, thereby improving the robustness of online monitoring.
[0004] The proposed main insulation online monitoring method does not need to be performed offline, and can monitor the health status of the motor insulation in real time. Once the insulation resistance drops to the alarm threshold, it will prompt the user to repair and replace it in advance to avoid potential safety hazards. Summary of the invention
[0005] In view of the above problems in the prior art, the present invention proposes an online monitoring method for early faults of the main insulation state of a three-phase asynchronous motor based on an improved SVR. By using the improved SVR to online predict the insulation resistance, the real-time monitoring of the motor health state is achieved, which provides a guarantee for the normal operation of the motor and has important practical significance for ensuring production safety.
[0006] The main function of the present invention is to monitor the insulation health status of the three-phase asynchronous motor online through insulation resistance. Traditional insulation resistance measurement methods mostly use offline diagnostic methods. During offline detection, the equipment needs to be shut down, which affects the economic benefits of production work. This method uses an online monitoring method for early faults of the main insulation status of a three-phase asynchronous motor based on an improved SVR for traditional insulation health monitoring, and predicts the insulation resistance value online in real time, and compares it with the set alarm threshold to achieve the purpose of real-time monitoring of the health status of the motor insulation and alarm.
[0007] The technical solution adopted by the present invention to solve the technical problem is: an online monitoring method for early fault of the main insulation state of a motor based on an improved SVR, comprising the following steps:
[0008] Obtain various data sets during the main insulation degradation process of the same model motor, and pre-process the data of each data set;
[0009] According to the preprocessed data set, the optimal parameters of the SVR model are obtained through the improved Grey Wolf algorithm, and an insulation resistance prediction model based on the improved SVR is constructed;
[0010] The sensor data is collected online to form various data sets and preprocessed, and the insulation resistance prediction value is obtained by improving the insulation resistance prediction model of SVR;
[0011] The predicted value is compared with the motor insulation resistance alarm threshold to achieve online monitoring of the motor main insulation status fault.
[0012] The data sets include:
[0013] The phase voltage V is collected online through the voltage sensor, current sensor and temperature sensor connected to the motor. a 、V b and V c , leakage current I and body temperature T;
[0014] According to the collected phase voltage Va 、V b and V c The common-mode voltage V is obtained by calculation CM , the common mode voltage V CM Perform FFT transformation on the drain current I to obtain V FFT and I FFT , and then get the equivalent capacitance C eq ;
[0015] The equivalent capacitance C eq , leakage current I FFT , body temperature T and insulation resistance R to construct various data sets.
[0016] The equivalent capacitance C eq Obtained through the following steps:
[0017]
[0018]
[0019] Where Z is the common mode impedance, V FFT and I FFT The common mode voltage V CM After Fourier transformation of the leakage current I, select the value of the inverter switching frequency connected to the motor at the first harmonic f, θ is the phase difference between the common-mode voltage and the common-mode leakage current at f, V a 、V b and V c is the phase voltage.
[0020] The construction of the insulation resistance prediction model based on the improved SVR comprises the following steps:
[0021] (1) Preprocess the data set to obtain the training set;
[0022] (2) Input the training set into the improved grey wolf algorithm and iterate;
[0023] (3) Use the training set to predict the SVR model, use the mean square error of the prediction results as the fitness function, and update α wolf, β wolf, and δ wolf;
[0024] (4) Obtain the optimal value, bring it into the SVR model to calculate the fitness, and return to step (3);
[0025] (5) Until the iteration is finished, the optimal parameters of the SVR model are output: penalty factor C and kernel parameter γ;
[0026] (6) Bring the optimal parameters into the SVR model to form a new SVR model;
[0027] (7) The training set is brought into the new SVR model for training to obtain the improved SVR insulation resistance prediction model.
[0028] The improved grey wolf algorithm comprises the following steps:
[0029] 1) Use improved Logistic to initialize the population: α wolf, β wolf and δ wolf are used to represent the penalty factor C and the kernel parameter γ. The position of the wolf represents the specific value of the penalty factor and the kernel parameter. Initialize the position X of α wolf, β wolf and δ wolf α X β and X δ , the specific formula is as follows:
[0030] X n+1 =sin(π((a-1)X n +(a-2)cos(πX n )))
[0031] Among them, X n is the position of the nth wolf;
[0032] 2) Tail prey: First, use the position formula D to update the distance between all gray wolves and their prey; use the distance formula X wolf Update the positions of all gray wolves:
[0033] D=|CX p (t)-X wolf (t)|
[0034] X wolf (t+1)=X P (t)-AD
[0035] Where D is the distance between the wolf and its prey; X p (t) represents the location of the prey; X wolf (t) represents the position of the gray wolf; t represents the number of iterations; A is a coefficient vector, and H inherits the position of the individual wolf by using an adaptive inheritance strategy. The calculation formulas are:
[0036] A=2ar1-a
[0037]
[0038]
[0039] Where r1 is a random vector in the range of (0, 1), H α , H β and H δ Represent the position inheritance capabilities of α wolf, β wolf and δ wolf respectively, a represents the convergence factor, t max Represents the maximum number of iterations;
[0040]
[0041] In the formula, represents the position of the wolf ω after iteration k+1;
[0042] 3) Hunting prey: Use the position formula D to update the distance between all gray wolves and their prey; use the distance formula X to update the positions of all gray wolves:
[0043] D α =|H α X α -X wolf |
[0044] D β =|H β X β -X wolf |
[0045] D δ =|H δ X δ -X wolf |
[0046] X1=X α -A1D α
[0047] X2=X β -A2D β
[0048] X3=X δ -A3D δ
[0049]
[0050] Where D α , D β , D δ Respectively represent the distances of α wolf, β wolf and δ wolf from their prey, X α X β and X δ represents the position of α wolf, β wolf and δ wolf; X1, X2 and X3 represent the displacement caused by α wolf, β wolf and δ wolf; X wolf (t+1) represents the position of ω wolf; A1, A2, A3 represent the coefficient vectors of α wolf, β wolf and δ wolf respectively;
[0051] 4) Attacking prey: ω wolf is in the optimal position X wolf (t+1) Attack.
[0052] The predicted value is compared with the motor insulation resistance alarm threshold, and if the predicted value is less than the set alarm threshold, a danger warning prompt is issued.
[0053] The online monitoring system for early fault of motor main insulation status based on improved SVR includes:
[0054] The preprocessing module is used to obtain various data sets during the main insulation degradation process of the same type of motor and preprocess the data of each data set;
[0055] A model building module is used to obtain the optimal parameters of the SVR model through the improved Grey Wolf Algorithm according to the preprocessed data set, and to build an insulation resistance prediction model based on the improved SVR;
[0056] The resistance prediction module is used to collect sensor data online to form various data sets and perform preprocessing, and obtain the insulation resistance prediction value by improving the insulation resistance prediction model of SVR;
[0057] The fault detection module is used to compare the predicted value with the motor insulation resistance alarm threshold to achieve online monitoring of the motor main insulation status fault.
[0058] The invention discloses an online monitoring device for early faults of main insulation state of a motor based on improved SVR, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to implement the online monitoring method for early faults of main insulation state of a motor based on improved SVR when executing the computer program.
[0059] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the online monitoring method for early faults of the main insulation state of a motor based on improved SVR is implemented.
[0060] The present invention has the following beneficial effects and advantages:
[0061] 1. Adopt non-intrusive monitoring. Since the data (phase voltage, leakage current, temperature) required to be collected by the present invention can be measured by voltage sensors, current transformers and temperature sensors, the online monitoring method for early faults of the main insulation state of three-phase asynchronous motors based on improved SVR can collect various data in the system without invading the system. The collection equipment is mainly composed of sensors, collection circuits, and microprocessors. The equipment does not need to be disassembled for installation, will not infringe on the privacy of users, is easy to install, and reduces costs.
[0062] 2. Improved prediction efficiency and accuracy. The online monitoring method for early faults of the main insulation state of the motor based on improved SVR uses machine learning methods and improves the SVR model. Compared with the traditional SVR model, it not only improves the prediction efficiency but also improves the prediction accuracy. It is not only accurate but also timely to monitor the equipment in the actual production process.
[0063] 3. Online monitoring of motor status. The online monitoring method for early faults of the motor main insulation status based on the improved SVR can collect a large amount of data online, realize online prediction of insulation resistance, and judge the health status of insulation by monitoring the value of insulation resistance. Without affecting normal operation, problems can be discovered and solved in time, thus reducing equipment downtime. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] Figure 1 It is a structural principle diagram of the present invention;
[0065] Figure 2 The hardware link diagram of the present invention;
[0066] Figure 3 It is a flow chart of constructing an insulation resistance prediction model based on SVR of the present invention;
[0067] Figure 4 It is a working flow chart of the method for online monitoring of early faults of main insulation state of three-phase asynchronous motor based on improved SVR of the present invention;
[0068] Figure 5 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION
[0069] The present invention is further described in detail below in conjunction with the embodiments.
[0070] The present invention provides an online monitoring method for early faults of the main insulation state of a three-phase asynchronous motor based on improved support vector regression (SVR), comprising the following steps: based on an accelerated degradation experiment or collecting historical data, obtaining a main insulation degradation data set of the three-phase asynchronous motor, including phase voltage, leakage current, body temperature and insulation resistance; preprocessing the data set, and constructing an insulation resistance prediction model based on the improved SVR; online collecting phase voltage and leakage current signals, calculating equivalent capacitance, inputting equivalent capacitance, leakage current and body temperature data into the insulation resistance prediction model to predict the insulation resistance of the motor online, and monitoring the health status of the equipment by real-time prediction of whether the insulation resistance value of the motor exceeds a set threshold.
[0071] like Figure 1 , Figure 2 As shown, the present invention, in terms of function, firstly uses the acquisition circuit to use three voltage sensors to measure the phase-to-ground voltage V of the three phases of the motor. a 、V b and V c , use the current transformer to measure the three-phase total leakage current I, and use the temperature sensor to measure the body temperature T. After collecting the data, the common mode voltage V is calculated CM, after Fourier transformation of the common mode voltage and leakage current, select the voltage value V at the first harmonic of the inverter switching frequency connected to the motor FFT and current value I FFT , and calculate the equivalent capacitance C by the formula eq . eq ,I FFT The insulation resistance and the body temperature T are input into the improved insulation resistance prediction model. After the predicted insulation resistance is obtained, it is compared with the set alarm threshold to conduct real-time evaluation and alarm on the health status of the motor main insulation.
[0072] The present invention relates to the following two key technologies:
[0073] 1) Online prediction of insulation resistance. The present invention first uses the instrument acquisition circuit to extract the voltage, current and temperature signals from the sensor data. The microprocessor pre-processes the data and calculates the equivalent capacitance C eq By adding C eq ,I FFT The body temperature T is input into the improved insulation resistance prediction model to realize the online prediction of insulation resistance and judge the health status of the motor by the change of insulation resistance value.
[0074] 2) Improve SVR technology. By using the improved Logistic chaos mapping algorithm and adaptive weight strategy to optimize the GWO algorithm and combining it with SVR, an SVR-based insulation resistance prediction model is constructed. This improves the prediction efficiency and accuracy of insulation resistance.
[0075] An online monitoring method for early faults of main insulation state of three-phase asynchronous motor based on improved SVR includes a sensor, a data acquisition circuit and a microprocessor; wherein the sensor is connected to the acquisition circuit, the acquisition circuit is connected to the microprocessor, and the microprocessor uses the acquisition circuit to collect sensor data in real time. First, the acquisition circuit is used to extract the phase voltage V in the sensor data. a 、V b and V c , leakage current I and body temperature T, calculate the equivalent capacitance C by the formula eq . eq ,I FFT The insulation resistance and the body temperature T are input into the improved insulation resistance prediction model. After the predicted insulation resistance is obtained, it is compared with the set alarm threshold to conduct real-time evaluation and alarm on the health status of the motor main insulation.
[0076] like Figure 3 , Figure 4 , Figure 5 As shown, the specific steps of the present invention are as follows:
[0077] The first step is to obtain various data sets of motor main insulation degradation:
[0078] By performing degradation experiments on the motor or collecting various data sets of the main insulation degradation process of the same model motor from historical motor degradation data, including the equivalent capacitance C eq (By phase voltage V a 、V b and V c and the leakage current I calculated), the leakage current I FFT , body temperature T and insulation resistance R.
[0079] The second step is data preprocessing:
[0080] Normalize the data sets collected in the first step.
[0081] The third step is to build an insulation resistance prediction model based on improved SVR:
[0082] The Grey Wolf Optimization (GWO) algorithm is optimized by using the improved Logistic chaotic mapping algorithm and the adaptive weight strategy, and combined with SVR to build an insulation resistance prediction model based on SVR. The specific steps are as follows:
[0083] 1-1) GWO is optimized using improved Logistic chaos mapping algorithm and adaptive weight strategy.
[0084] 1-2) Use the improved GWO combined with SVR, and use GWO to automatically find the optimal values of the SVR related parameters penalty factor C and kernel parameter γ.
[0085] The detailed steps of 1-1) are as follows:
[0086] (1) Use improved Logistic initialization population: α wolf, β wolf and δ wolf are used to represent the penalty factor C and the kernel parameter γ, that is, the position of the wolf represents the specific value of the penalty factor and the kernel parameter. Initialize the position X of α wolf, β wolf and δ wolf α X β and X δ , the improved Logistic specific formula is as follows:
[0087] X n+1 =sin(π((a-1)X n +(a-2)cos(πX n )))
[0088] Among them, X n is the position of the nth wolf;
[0089] (2) Tail prey: First, use the position formula D to update the distance between all gray wolves and their prey; use the distance formula Xwolf Update the positions of all gray wolves. The formula is as follows:
[0090] D=|CX p (t)-X wolf (t)|
[0091] X wolf (t+1)=X P (t)-AD
[0092] Where D is the distance between the wolf and its prey; X p (t) indicates the location of the prey;
[0093] X wolf (t) represents the position of the gray wolf; t represents the number of iterations; A is a coefficient vector, and H inherits the position of the individual wolf by using an adaptive inheritance strategy. The calculation formulas are:
[0094] A=2ar1-a
[0095]
[0096]
[0097] Where r1 is a random vector in the range of (0, 1), H α , H β and H δ Represent the position inheritance capabilities of α wolf, β wolf and δ wolf respectively, a represents the convergence factor, t max Represents the maximum number of iterations;
[0098]
[0099] Where: Represents the position of ω wolf after iteration k+1.
[0100] (3) Hunting prey: Use the position formula D to update the distance between all gray wolves and their prey; use the distance formula X to update the positions of all gray wolves. The formula is as follows:
[0101] D α =|H α X α -X wolf |
[0102] D β =|H β X β -X wolf |
[0103] D δ =|H δ X δ -Xwolf |
[0104] X1=X α -A1D α
[0105] X2=X β -A2D β
[0106] X3=X δ -A3D δ
[0107]
[0108] Where D α , D β , D δ Respectively represent the distances of α wolf, β wolf and δ wolf from their prey, X α X β and X δ represents the position of α wolf, β wolf and δ wolf; X1 X2 and X3 represent the displacement caused by α wolf, β wolf and δ wolf; X wolf (t+1) represents the position of ω wolf; A1, A2, A3 represent the coefficient vectors of α wolf, β wolf and δ wolf respectively;
[0109] (4) Attacking prey: ω wolf is in the optimal position X wolf (t+1) Attack.
[0110] The detailed steps of steps 1-2) are as follows:
[0111] (1) Data processing: The data set is first preprocessed, and the processed data is the training set.
[0112] (2) Input the improved GWO algorithm: Input the training set into the improved GWO algorithm for iteration.
[0113] (3) Calculate fitness: Use the training set to predict SVR, use the mean squared error (MSE) of the prediction results as the fitness function, and update α wolf, β wolf and δ wolf.
[0114] (4) Obtain the optimal value, bring it into the SVR model to calculate the fitness, and repeat step 3.
[0115] (5) Until the iteration ends, the optimal SVR parameters are output.
[0116] (6) Bring the optimal parameters into SVR to form a new SVR prediction model.
[0117] (7) The training set is fed into the new SVR model for training.
[0118] Step 4: Collect sensor data online:
[0119] The phase voltage V is collected online through the voltage sensor, current sensor and temperature sensor connected to the motor. a 、V b and V c , leakage current I and body temperature T.
[0120] Step 5: Calculate the equivalent capacitance using FFT transformation:
[0121] The phase voltage V collected in step 4) a 、V b and V c The common-mode voltage V is obtained by calculation CM , the common mode voltage V CM Perform FFT transformation on the drain current I to obtain V FFT and I FFT , calculate the common mode voltage V by the formula CM And the equivalent capacitance C eq , the calculation formula is as follows:
[0122]
[0123]
[0124] Where Z is the common mode impedance, V FFT and I FFT After FFT of the common-mode voltage and common-mode leakage current respectively, the value at the first harmonic f of the inverter switching frequency connected to the motor is selected, and θ is the phase difference between the common-mode voltage and the common-mode leakage current at f.
[0125] Step 6: Input prediction model:
[0126] The C obtained in step 5 eq ,I FFT and the body temperature T are input into the insulation resistance prediction model constructed in step 3).
[0127] Step 7: Output the insulation resistance prediction results:
[0128] The real-time predicted insulation resistance is obtained by improving the SVR prediction model.
[0129] Step 8: Compare the predicted value with the motor insulation resistance alarm threshold:
[0130] By comparing the predicted value output in step 7 with the set alarm threshold, the health status of the motor main insulation is evaluated and alarmed in real time. The alarm threshold can be set by the user according to the insulation level of the motor, historical experience, etc.
Claims
1. An online monitoring method for early fault of main insulation state of a motor based on improved SVR is characterized by: The following steps are involved: Obtain various data sets during the main insulation degradation process of the same model motor, and pre-process the data of each data set; According to the preprocessed data set, the optimal parameters of the SVR model are obtained through the improved Grey Wolf Algorithm, and an insulation resistance prediction model based on the improved SVR is constructed; The sensor data is collected online to form various data sets and preprocessed, and the insulation resistance prediction value is obtained by improving the insulation resistance prediction model of SVR; The predicted value is compared with the motor insulation resistance alarm threshold to achieve online monitoring of the motor main insulation status fault.
2. The method for online monitoring of early faults of main insulation state of a motor based on improved SVR according to claim 1 is characterized in that: The data sets include: The phase voltage V is collected online through the voltage sensor, current sensor and temperature sensor connected to the motor. a 、V b and V c , leakage current I and body temperature T; According to the collected phase voltage V a 、V b and V c The common-mode voltage V is obtained by calculation CM , the common mode voltage V CM Perform FFT transformation on the drain current I to obtain V FFT and I FFT , and then get the equivalent capacitance C eq ; The equivalent capacitance C eq , leakage current I FFT , body temperature T and insulation resistance R to construct various data sets.
3. The method for online monitoring of early faults of main insulation state of a motor based on improved SVR according to claim 2 is characterized in that: The equivalent capacitance C eq Obtained through the following steps: Where Z is the common mode impedance, V FFT and I FFT They are the common mode voltage V CM After Fourier transformation of the leakage current I, select the value of the inverter switching frequency connected to the motor at the first harmonic f, θ is the phase difference between the common-mode voltage and the common-mode leakage current at f, V a 、V b and V c is the phase voltage.
4. The method for online monitoring of early faults of main insulation state of a motor based on improved SVR according to claim 1 is characterized in that: The construction of the insulation resistance prediction model based on the improved SVR comprises the following steps: (1) Preprocess the data set to obtain the training set; (2) Input the training set into the improved grey wolf algorithm and iterate; (3) Use the training set to predict the SVR model, use the mean square error of the prediction results as the fitness function, and update α wolf, β wolf, and δ wolf; (4) Obtain the optimal value, bring it into the SVR model to calculate the fitness, and return to step (3); (5) Until the iteration is finished, the optimal parameters of the SVR model are output: penalty factor C and kernel parameter γ; (6) Bring the optimal parameters into the SVR model to form a new SVR model; (7) The training set is brought into the new SVR model for training to obtain the improved SVR insulation resistance prediction model.
5. The method for online monitoring of early faults of main insulation state of a motor based on improved SVR according to claim 1 or 4, characterized in that: The improved grey wolf algorithm comprises the following steps: 1) Use improved Logistic to initialize the population: α wolf, β wolf and δ wolf are used to represent the penalty factor C and the kernel parameter γ. The position of the wolf represents the specific value of the penalty factor and the kernel parameter. Initialize the position X of α wolf, β wolf and δ wolf α X β and X δ , the specific formula is as follows: X n+1 =sin(π((a-1)X n +(a-2)cos(πX n ))) Among them, X n is the position of the nth wolf; 2) Tail prey: First, use the position formula D to update the distance between all gray wolves and their prey; use the distance formula X wolf Update the positions of all gray wolves: D=|CX p (t)-X wolf (t)| X wolf (t+1)=X P (t)-AD Where D is the distance between the wolf and its prey; X p (t) indicates the location of the prey; X wolf (t) represents the position of the gray wolf; t represents the number of iterations; A is a coefficient vector, and H inherits the position of the individual wolf by using an adaptive inheritance strategy. The calculation formulas are: A=2ar1-a Where r1 is a random vector in the range of (0, 1), H α , H β and H δ Represent the position inheritance capabilities of α wolf, β wolf and δ wolf respectively, a represents the convergence factor, t max Represents the maximum number of iterations; Where, X wolf k+1 represents the position of the wolf ω after iteration k+1; 3) Hunting prey: Use the position formula D to update the distance between all gray wolves and their prey; use the distance formula X to update the positions of all gray wolves: D α =|H α X α -X wolf | D β =|H β X β -X wolf | D δ =|H δ X δ -X wolf | X1=X α -A1D α X2=X β -A2D β <h2 style=";text-align:left;direction:ltr">X3=X<h2 style=";text-align:left;direction:ltr"> δ <h2 style=";text-align:left;direction:ltr"> -A3D<h2 style=";text-align:left;direction:ltr"> δ Where D α , D β , D δ Respectively represent the distances of α wolf, β wolf and δ wolf from their prey, X α X β and X δ represents the position of α wolf, β wolf and δ wolf; X1, X2 and X3 represent the displacement caused by α wolf, β wolf and δ wolf; X wolf (t+1) represents the position of ω wolf; A1, A2, A3 represent the coefficient vectors of α wolf, β wolf and δ wolf respectively; 4) Attacking prey: ω wolf is in the optimal position X wolf (t+1) Attack.
6. The method for online monitoring of early faults of main insulation state of a motor based on improved SVR according to claim 1 is characterized in that: The predicted value is compared with the motor insulation resistance alarm threshold, and if the predicted value is less than the set alarm threshold, a danger warning prompt is issued.
7. The online monitoring system for early fault of main insulation state of motor based on improved SVR is characterized by: include: The preprocessing module is used to obtain various data sets during the main insulation degradation process of the same type of motor and preprocess the data of each data set; A model building module is used to obtain the optimal parameters of the SVR model through the improved Grey Wolf Algorithm according to the preprocessed data set, and to build an insulation resistance prediction model based on the improved SVR; The resistance prediction module is used to collect sensor data online to form various data sets and perform preprocessing, and obtain the insulation resistance prediction value by improving the insulation resistance prediction model of SVR; The fault detection module is used to compare the predicted value with the motor insulation resistance alarm threshold to achieve online monitoring of the motor main insulation status fault.
8. The online monitoring device for early fault of main insulation state of motor based on improved SVR is characterized by: It comprises a memory and a processor; the memory is used to store a computer program; the processor is used to implement the online monitoring method for early fault of the main insulation state of a motor based on the improved SVR as described in any one of claims 1 to 6 when executing the computer program.
9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the online monitoring method for early faults of the main insulation state of a motor based on the improved SVR as described in any one of claims 1 to 6 is implemented.
Citation Information
Cited By
Vehicle control method and device
CN119858450A
System and method for evaluating insulating property of automobile motor
CN120161338A
Variable frequency motor stator insulation monitoring method based on resonant frequency multi-dimensional characteristic cloud picture
CN120539546A
Traction motor fault on-line monitoring method and system based on leakage current impedance characteristic
CN121276326A
Method and system for online monitoring of traction motor faults based on leakage current impedance characteristics
CN121276326B
