Search

CN-115825732-B - Intelligent diagnosis method for open-circuit faults of characteristic-associated permanent magnet synchronous motor driving system

CN115825732BCN 115825732 BCN115825732 BCN 115825732BCN-115825732-B

Abstract

The invention discloses an intelligent diagnosis method for open-circuit faults of a permanent magnet synchronous motor driving system with characteristic association, which comprises the steps of firstly, combining electrical characteristic analysis to obtain torque signals with higher distinction degree as input of a neural network, collecting torque signals under a specific position angle and normalizing the torque signals to [ -1,1] to form a training set, secondly, defining fault characteristics to describe open-phase and switching tube faults, training an improved extreme learning machine to predict the defined fault characteristics, and finally, comparing the predicted fault characteristics with target fault characteristics under each fault through design reasoning rules to finally diagnose fault types. Compared with the similar method, the method can be used for directly training the neural network through a large amount of data to predict the final fault category, the method can be used for effectively reducing the scale of the neural network and the required data quantity, and the problem that the existing similar method cannot be used for processing unknown faults can be solved.

Inventors

  • MAO YAO
  • Jin Luhan
  • WANG XUEQING
  • LU LINLIN

Assignees

  • 中国科学院光电技术研究所
  • 四川大学

Dates

Publication Date
20260512
Application Date
20221206

Claims (7)

  1. 1. The intelligent diagnosis method for the open-circuit fault of the permanent magnet synchronous motor driving system with the characteristic association is characterized by comprising the following steps of: Step 1, acquiring a torque signal under a specific position angle as input of a neural network, normalizing to [ -1,1], and constructing a training set; Step 2, defining fault characteristics to describe open-phase faults and open-circuit faults of a switching tube, and training an extreme learning machine to predict the fault characteristics; step 3, binarizing the predicted fault characteristics to obtain predicted fault characteristic vectors; step 4, finally determining fault categories by comparing the predicted fault feature vectors with target feature vectors under all faults; theoretical analysis proves that the torque signal in the step 1 can effectively distinguish all open-phase faults and open-circuit faults of the switching tube, and for the open-phase faults of the phase A and the open-circuit faults of the switching tube, the torque signal after different faults occur approximately has the following expression by combining fault mechanism and electrical characteristic analysis: Wherein, the 、 、 Respectively representing torque output signals after phase A faults, upper switching tube open circuit and lower switching tube open circuit, N p is pole pair number of the motor, psi f is amplitude of rotor flux linkage, C represents estimation coefficient, theta e is electrical angle of the rotor, Is the torque output value of the motor in the normal state.
  2. 2. The intelligent diagnosis method for open-circuit faults of a characteristic-associated permanent magnet synchronous motor driving system according to claim 1, wherein normalization to [ -1,1] in the step 1 is achieved by the following function: Wherein x and x normal are the original input vector and the normalized input vector, respectively, min (x) represents the minimum value in the input vector x, and max (x) represents the maximum value in the input vector x.
  3. 3. The intelligent diagnosis method for open-circuit faults of the permanent magnet synchronous motor driving system with characteristic association according to claim 1, wherein the fault characteristics in the step 2 are defined manually and are used for describing each fault, so that the original multi-classification problem of fault types is converted into the bi-classification problem of single fault characteristics, and the size of a neural network is reduced.
  4. 4. The intelligent diagnosis method for the open-circuit fault of the characteristic-associated permanent magnet synchronous motor driving system according to claim 1 is characterized in that the fault characteristic in the step 2 can introduce fault information of unknown working conditions into the learning of a neural network to realize the fault diagnosis under the unknown working conditions.
  5. 5. The intelligent diagnosis method for open-circuit faults of a characteristic-associated permanent magnet synchronous motor driving system according to claim 1, wherein the extreme learning machine in the step 2 is an improved integrated parallel extreme learning machine and is used for generating a mapping relation from an input torque signal to an output fault characteristic.
  6. 6. The intelligent diagnosis method for open-circuit faults of the characteristic-associated permanent magnet synchronous motor driving system according to claim 1, wherein the binarization processing in the step 3 is realized by the following functions: wherein f i and f i-b respectively represent the original predicted fault characteristics of the neural network and the predicted fault characteristics after binarization processing, and f th is a preset threshold.
  7. 7. The intelligent diagnosis method for open-circuit faults of a feature-associated permanent magnet synchronous motor driving system according to claim 1, wherein in the step 4, the predicted fault feature vector and the target feature vector under each fault are compared to finally determine a fault class, and the fault class corresponding to the target feature vector which is completely consistent with the predicted fault feature vector is selected as the final output, and when no target feature vector is identical with the predicted fault feature vector, the unknown fault is indicated to occur at the moment.

Description

Intelligent diagnosis method for open-circuit faults of characteristic-associated permanent magnet synchronous motor driving system Technical Field The invention relates to the technical field of state monitoring and fault diagnosis of motor drive systems, in particular to an intelligent diagnosis method for open-circuit faults of a permanent magnet synchronous motor drive system with characteristic association. Background Because the permanent magnet synchronous motor has the advantages of simple structure, reliable operation, high efficiency, flexible control and the like, the permanent magnet synchronous motor is widely applied to the fields with high requirements on reliability and power density, such as aerospace, electric automobiles, robots, ship propulsion and the like. However, the application environment of the practical system is very complex, and the permanent magnet synchronous motor driving system is required to resist the influence of internal disturbance and external disturbance on the system at the same time, so that the possible types of faults are numerous. The open-circuit fault is the most common electrical fault in the permanent magnet synchronous motor driving system, and comprises two major types of open-phase faults and open-circuit faults of a switching tube. These two types of faults usually do not immediately cause the downtime of the system, but generate large torque pulsation and mechanical vibration, which seriously affect the operation safety of the system. Therefore, the method has great significance for accurately diagnosing the open circuit fault of the permanent magnet synchronous motor driving system. In recent years, with the increasing computing power of embedded devices, data-based intelligent diagnostic methods are increasingly favored by more researchers due to their superior performance. However, existing intelligent data-based diagnostic methods often require the collection of large amounts of fault data from the actual system for training of large neural networks. Large-scale data acquisition can cause irreversible damage to the actual system, severely affecting the service life of the system. In addition, the method has the problems of high algorithm complexity, incapability of processing unknown faults and the like, and is difficult to apply to an actual system. Disclosure of Invention The invention aims to overcome the defects of the existing similar method in the background art and provides an intelligent diagnosis method for the open-circuit fault of a characteristic-associated permanent magnet synchronous motor driving system. The technical scheme adopted by the invention is that the intelligent diagnosis method for the open-circuit fault of the permanent magnet synchronous motor driving system with characteristic association comprises the following steps: Step 1, acquiring a torque signal under a specific position angle as input of a neural network, normalizing to [ -1,1], and constructing a training set; Step 2, defining fault characteristics to describe open-phase faults and open-circuit faults of a switching tube, and training an extreme learning machine to predict the fault characteristics; step 3, binarizing the predicted fault characteristics to obtain predicted fault characteristic vectors; and 4, finally determining the fault category by comparing the predicted fault characteristic vector with the target characteristic vector under each fault. Further, theoretical analysis proves that the torque signal in the step 1 can effectively distinguish all open-phase faults and open-circuit faults of the switching tube, and for the open-phase faults of the phase A and the open-circuit faults of the switching tube, the torque signal after different faults occur approximately has the following expression by combining fault mechanism and electrical characteristic analysis: wherein T e_OPF-A, The torque output signals after the phase A fault, the upper switching tube open circuit and the lower switching tube open circuit are respectively represented, N p is the pole pair number of the motor, ψ f is the amplitude of the rotor flux linkage, C represents the estimation coefficient, θ e is the electrical angle of the rotor, and T e* is the torque output value of the motor in the normal state. Further, the normalization to [ -1,1] in step 1 is achieved by the following function: Wherein x and x normal are the original input vector and the normalized input vector, respectively, and min (x) and max (x) represent the minimum value and the maximum value in the input vector x, respectively. Further, the fault characteristics described in the step 2 are defined manually, and are used for describing each fault, so that the original multi-classification problem of the fault type can be converted into the two-classification problem of the single fault characteristic, and the scale of the neural network is reduced. Further, the fault characteristics in the step 2 can introduc