TWI426277B - Motor test system and method thereof - Google Patents

Description

Motor detection system and method thereof

The present invention relates to a detection system, and more particularly to a motor detection system.

Electric motors have been widely used in various applications. Among them, with the problem of global warming and oil depletion, electric vehicles are a new trend in the future. Among them, the wheel-type motor can be directly driven in the wheel, and it is not necessary to decelerate like a conventional engine vehicle. The transmission has many advantages such as simple structure and low cost.

However, the motor may generate various abnormal states and losses with aging or external environmental factors. For example, a hub type motor disposed in a wheel may have an abnormal state such as insulation damage or bearing damage due to aging or external environmental factors. If the abnormal state of the motor can be found immediately, the subsequent serious faults can be avoided, thereby saving a lot of maintenance costs or parts replacement costs.

Conventional motor detection techniques usually use a vibration sensor or a temperature sensor to check whether the motor is in an abnormal state by detecting vibration and temperature during motor operation. However, the above-mentioned conventional motor detection technology not only needs to increase the additional cost of the sensor and the measuring instrument, but also often causes measurement error due to sensor aging or sensor position shift. The abnormal state of the motor cannot be recognized.

In view of the shortcomings of the prior art described above, it is necessary to provide a motor detection system that has better recognition capabilities for abnormal motors.

The problem to be solved by the present invention is to provide a motor detection system that has better identification capabilities for abnormal motors.

In order to solve the above problems, an embodiment of the present invention provides a motor detection system including a current signal acquisition unit, a data conversion unit, and an automatic identification unit. The current signal acquisition unit is configured to capture a current signal of a motor; the data conversion unit is configured to generate a characteristic data from the current signal; and the automatic identification unit is configured to compare the characteristic data with an abnormal feature data to identify One of the states of the motor.

Another embodiment of the present invention provides a motor detection method including the following steps. First, a current signal of a motor is captured; then, the current signal is converted to generate a feature data; and then the feature data is compared with an abnormal feature data to identify a state of the motor.

The motor detecting system and the detecting method of the invention directly recognize the state of the motor by the current signal of the motor, and do not need to additionally provide a vibration sensor or a temperature sensor, so that the sensor does not age or the sensor position The motor detection system and the detection method of the present invention have better identification ability for the abnormal motor due to the offset and the measurement error. At the same time, the motor detection system and the detection method of the present invention are low in cost and can be generally applied to various uses.

100‧‧‧Motor system

110‧‧‧Electric motor

120‧‧‧ Controller

130‧‧‧Power supply

200‧‧‧Detection system

210‧‧‧current signal acquisition unit

220‧‧‧Data Conversion Unit

230‧‧‧Automatic identification unit

300‧‧‧Motor testing method

310‧‧‧Receiving the current signal of the motor

320‧‧‧Data conversion of current signals to generate characteristic data

330‧‧‧Comparing the characteristic data with the abnormal feature data to identify the state of the motor

340‧‧‧ Establishing anomalous characteristics data

341‧‧‧Drive abnormal motor

342‧‧‧Draw the current signal of abnormal motor

343‧‧‧Data conversion of current signals to generate anomalous characteristics

344‧‧‧Storage anomaly characteristics

The first figure shows a schematic diagram of a motor detection system in accordance with a preferred embodiment of the present invention.

The second figure shows a schematic flow chart of an abnormality identification method for a motor according to another preferred embodiment of the present invention.

The third figure shows a schematic diagram of the flow of the motor detection method applied to the hub type motor.

Some embodiments of the invention are described in detail below. However, the present invention may be widely practiced in other embodiments than the following description, and the scope of the present invention is not limited by the examples, which are subject to the scope of the following patents. Further, in order to provide a clearer description and a better understanding of the present invention, the various parts of the drawings are not drawn according to their relative dimensions, and some dimensions have been exaggerated compared to other related dimensions; the irrelevant details are not fully drawn. Out, in order to make the schema simple.

The first figure shows a schematic diagram of a motor detection system 200 in accordance with a preferred embodiment of the present invention, wherein the motor detection system 200 is used to detect the motor system 100.

The motor system 100 includes a motor 110, a controller 120, and a power source 130. The controller 120 is electrically connected to the power source 130. The motor 110 is electrically connected to the controller 120, and the controller 120 is used to drive the motor 110. In the present embodiment, the motor 110 may be a DC brushless motor, such as a hub type motor. However, it is not limited thereto, and the motor 110 may be other types of motors.

The motor detection system 200 includes a current signal acquisition unit 210 and a data The conversion unit 220 and an automatic identification unit 230. The current signal capturing unit 210 is configured to capture a current signal of the motor 110. The data conversion unit 220 is configured to perform data conversion on the current signal to generate a feature data. The automatic identification unit 230 is configured to use the feature data with an abnormality. The feature data is compared to identify the operating state of the motor 110.

In this embodiment, the data conversion unit 220 performs data conversion on the captured current signal by using a Hilbert-yellow conversion method to generate feature data. The Hilbert-Yellow conversion method is mainly composed of empirical modal analysis and Hilbert conversion. It not only displays the signal conversion frequency correctly on the spectrum, but also applies to nonlinear and non-steady-state signals. Therefore, it can be significantly improved. Noise problems due to vibration signals. In this embodiment, the data conversion unit 220 performs data conversion on the captured current signal by using the Hilbert-yellow conversion method, but is not limited thereto, and the data conversion unit 220 may use other data conversion methods. .

In this embodiment, the automatic identification unit 230 compares the feature data with the abnormal feature data by using a reverse neural network to identify the state of the motor 110. Among them, the inverted neural network is a supervised neural network with higher accuracy than the unsupervised neural network. The automatic identification unit 230 first trains the abnormal characteristic data of the abnormal motor, and the obtained weight value of the inverted neural network can be used by other automatic identification units 230. In this embodiment, the automatic identification unit 230 compares the feature data with the abnormal feature data by using the reverse neural network to identify the state of the motor 110, but not limited thereto, the automatic identification unit 230 may also Use other identification methods.

With the motor detection system 200 of the present invention, the state of the motor 110 is directly recognized by the current signal of the motor 110, and no vibration sensor or temperature sensor needs to be provided, so that the sensor is not aged or the sensor position is not caused. The offset detection produces a measurement error, and the motor detection system 200 of the present invention has a better ability to recognize an abnormal motor. with At the time, the motor detecting system 200 of the present invention is low in cost and can be generally applied to various uses.

The second figure shows a schematic flow chart of a motor detecting method 300 according to another preferred embodiment of the present invention. The motor detection method 300 includes the following steps. Referring to the first and second figures, first, step 310 is performed, and a current signal of the motor 110 is captured by the current signal capturing unit 210. Then, step 320 is performed, and the current is converted by the data conversion unit 220. The signal is converted to generate a feature data. Then, step 330 is performed, and the feature data is compared with an abnormal feature data by the automatic identification unit 230 to identify the operating state of the motor 110. In the present embodiment, the motor 110 may be a DC brushless motor, such as a hub type motor. However, it is not limited thereto, and the motor 110 may be other types of motors.

In this embodiment, the motor detection method 300 may further include the step 340 of establishing the abnormal feature data. First, step 341 is performed to drive an abnormal motor; then, in step 342, a current signal of the abnormal motor is captured by the current signal capturing unit 210; then, step 343 is performed, by means of the data conversion unit 220, The current signal performs data conversion to generate abnormal feature data. Finally, step 344 is performed to store the abnormal feature data in the automatic identification unit 230.

In this embodiment, the data conversion unit 220 performs data conversion on the captured current signal by using a Hilbert-yellow conversion method to generate feature data. The Hilbert-Yellow conversion method is mainly composed of empirical modal analysis and Hilbert conversion. It not only displays the signal conversion frequency correctly on the spectrum, but also applies to nonlinear and non-steady-state signals. Therefore, it can be significantly improved. Noise problems due to vibration signals. Although, in this embodiment, the data conversion unit 220 takes Hill The Bo-Yellow conversion method performs data conversion on the extracted current signal, but is not limited thereto. The data conversion unit 220 can also use other data conversion methods.

In this embodiment, the automatic identification unit 230 compares the feature data with the abnormal feature data by using a reverse neural network to identify the state of the motor 110. Among them, the inverted neural network is a supervised neural network with higher accuracy than the unsupervised neural network. The automatic identification unit 230 first trains the abnormal characteristic data of the abnormal motor, and the obtained weight value of the inverted neural network can be used by other automatic identification units 230. In this embodiment, the automatic identification unit 230 compares the feature data with the abnormal feature data by using the reverse neural network to identify the state of the motor 110, but not limited thereto, the automatic identification unit 230 may also Use other identification methods.

The third diagram shows a flow diagram of the motor detection method 300 applied to a hub-type motor. In this embodiment, the data conversion unit 220 performs data conversion on the current signal of the hub type motor by using the Hilbert-yellow conversion method, thereby generating characteristic data F1-F50. Finally, the feature data F1-F50 is input to the inverted neural network of the automatic identification unit 230 for training. The automatic identification unit 230 that completes the training can recognize the operating state of the hub type motor.

As shown in the second and third figures, a step 340 of establishing abnormal feature data is performed prior to identifying the operational state of the hub motor. First, step 341 is performed to drive the abnormal hub motor; then, in step 342, the current signal extraction unit 210 is used to capture the current signal of the abnormal hub motor; then, step 343 is performed, and the data conversion unit 220 is performed by Hill. The Bosch-Yellow Conversion Method (HHT) converts the current signal of the abnormal hub motor to obtain a Hilbert conversion matrix (HT-matrix). Among them, the Hilbert transform matrix (HT-matrix) consists of time, frequency and amplitude. Obtain the maximum, average, standard deviation, root mean square, and energy time characteristic of each time point from the time axis of the Hilbert transformation matrix. Tmax, Tmean, Tstd, Trms, and Te; the frequency characteristic of each frequency point is obtained from the frequency axis of the Hilbert transformation matrix, and the frequency characteristic curve of the mean value, the standard deviation, the root mean square, and the energy is Fmax , Fmean, Fstd, Frms and Fe. Then obtain the five characteristics of the maximum value, average value, standard deviation value, root mean square value and energy value of the above ten characteristic curves, and finally obtain the abnormal feature data composed of the characteristic values F1-F50. Then, step 344 is performed to store the abnormal feature data described above in the automatic identification unit 230.

Through the above steps, the characteristic data of various operating states of the hub type motor is input to the inverted transmission neural network of the automatic identification unit 230 for training. The automatic identification unit 230 that completes the training can recognize the operating state of the hub type motor.

Table 1 shows the results of the hub type motor detection by the automatic identification unit 230 that completed the training. As shown in Table 1, a total of 100 hub-type motors were tested, including 20 normal hub motors, 20 single-slot insulated broken hub motors, 20 three-slot insulated broken hub motors, and 20 half-turn insulation. Destroyed hub-type motor and 20 full-circle insulation-damaged hub motors. As shown in Table 1, the total identification rate of dielectric breakdown is up to 96%.

Table 2 shows the results of the hub type motor detection by the automatic identification unit 230 that completed the training. As shown in Table 2, a total of 60 hub motors were tested, including 20 normal hub motors, 20 bearing perforated hub motors, and 20 hub motors with damaged covers. As shown in Table 2, the total identification rate of bearing damage can reach 98.33%.

With the motor detecting method 300 of the present invention, the state of the motor is directly recognized by the current signal of the motor, and there is no need to provide a vibration sensor or a temperature sensor, so that the sensor is not aged or the sensor position is shifted. The motor detection method 300 of the present invention has a better ability to recognize abnormal motors. At the same time, the motor detecting method 300 of the present invention is low in cost and can be generally applied to various uses.

The embodiments of the present invention are merely illustrative of the technical spirit and characteristics of the present invention, and the objects of the present invention can be understood by those skilled in the art, and the scope of the present invention cannot be limited thereto. All other equivalents and modifications of the inventions which are made without departing from the spirit of the invention are intended to be included within the scope of the invention.

100‧‧‧Motor system

110‧‧‧Electric motor

120‧‧‧ Controller

130‧‧‧Power supply

200‧‧‧Motor Detection System

210‧‧‧current signal acquisition unit

220‧‧‧Data Conversion Unit

230‧‧‧Automatic identification unit

Claims (5)

A motor detection system includes: a current signal acquisition unit for capturing a current signal of a motor; and a data conversion unit for converting data of the current signal by a Hilbert-yellow conversion method, thereby Generating a feature data; and an automatic identification unit includes an abnormal feature data, wherein the abnormal feature data is converted to an abnormal current signal of the plurality of abnormal motors by the Hilbert-yellow conversion method to generate a The abnormal feature data is further transmitted through the neural network training through one of the automatic identification units, and the automatic identification unit compares the feature data with the abnormal feature data to identify a state of the motor. The motor detection system of claim 1, wherein the motor comprises a brushless motor. The motor detecting system of claim 2, wherein the brushless DC motor comprises a hub type motor. An abnormality identification method for a motor comprises: driving a plurality of abnormal motors; extracting an abnormal current signal of the abnormal motor; and converting the abnormal current signal by a Hilbert-yellow conversion method, thereby generating the abnormal feature Data again; and training and storing the abnormal feature data via a reverse neural network. Taking a current signal of a motor; converting the current signal by the Hilbert-yellow conversion method to generate a feature And comparing the characteristic data with an abnormal feature data to identify a state of the motor. The method for identifying an abnormality of a motor according to claim 4, wherein the motor comprises a brushless motor.