Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities

Many companies in the industrial equipment market hope to use artificial intelligence and machine learning technology to help them minimize unexpected service interruptions and prevent their products, production lines and services from being affected. The reason for this idea is obvious: reducing unplanned downtime can improve operational efficiency and maximize benefits. A research report released by market analysis company Aberdeen in 2016 showed that the average cost of a company’s one-hour shutdown was as high as US$260,0001. OEMs can help customers implement predictive maintenance (PDM) systems by adding intelligent functions for measuring and analyzing performance data to industrial systems to identify and replace faulty systems

Lattice White Paper

introduction

Many companies in the industrial equipment market hope to use artificial intelligence and machine learning technology to help them minimize unexpected service interruptions and prevent their products, production lines and services from being affected. The reason for this idea is obvious: reducing unplanned downtime can improve operational efficiency and maximize benefits. A research report released by market analysis company Aberdeen in 2016 showed that the average cost of a company’s one-hour shutdown was as high as US$260,0001. OEMs can help customers implement predictive maintenance (PDM) systems by adding intelligent functions for measuring and analyzing performance data to industrial systems, so as to identify and replace faulty system components (such as motors used in industrial robots) to prevent them from malfunctioning. Discontinue production.

To help industrial equipment OEMs implement PDM functions in their products, Lattice Semiconductor has developed a collection of Lattice AutomateTM solutions for industrial automation systems. Lattice offers a variety of low-power FPGAs as a reprogrammable chip that can perform data processing or co-processing functions to build AI/ML inference models for PDM applications. In order to simplify and speed up the development of PDM systems based on Lattice FPGAs, Automate includes software tools, industrial IP cores, modular hardware development boards, software programmable reference designs and demonstrations. It is easy to design multi-functions with PDM functions and expandability. Channel motor control application. Figure 1 shows a motor control system with predictive maintenance capabilities based on the Lattice Automate solution collection design.

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 1. Motor control system with predictive maintenance function designed based on the Lattice Automate solution set, which can be used in industrial automation applications

Automate includes a PDM multi-channel motor control reference design, using the industry’s common motor current signal analysis (MCSA) technology. In Lattice’s solution, the Clark transformation converts the current from a three-phase motor into two signals. The converted current becomes α current and β current. For a healthy motor in normal operation, α current and β current are separated by 90 degrees. In the plane of the xy coordinate system, the trajectory of the point forms a circle. In the following, we will show various trajectory circle deformations caused by current or load imbalance.

This article uses the sensorless space vector pulse width modulation (SV_PWM) technology implemented in FPGA RTL to drive a three-phase brushless DC (BLDC) motor. The SV_PWM control signal drives the Trenz TEP0002 motor drive board. The development board implements a Hall current sensor and is connected to the motor to detect the motor winding current. The on-board ADC digitizes the output of the Hall current sensor, so this reference design can read and control the ADC used for motor control and PDM. The current is sampled at a rate of 0.8 MS/s per second.

Use the Clark transformation (Equation 1) to convert the three-phase (A, B, and C) currents (Figure 2) into α and β currents, as shown in Figure 3.

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 2. Three-phase motor current (IA, IB, and IC)

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 3. Clarke transform output Iα and Iβ

Observing Iα and Iβ, we can find that they are similar to the cos and sin functions. In fact, when they draw a function image on the xy coordinate plane, the result is a circle (Figure 4).

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 4. XY plane images of Iα and Iβ

The reference design can collect a snapshot of the motor winding current for a fixed number of shaft rotations (the default is a snapshot of 50 revolutions) or the user can select a snapshot of the winding current over a longer period of time. Before applying the Clark transform, the signal processing of the motor current includes peak detection and normalization and a moving average filter. This adaptive function is suitable for various motors and PDMs with various power consumption levels.

The PDM solution of the Lattice Automate solution set includes a proprietary algorithm that folds the circle data (as shown in Figure 6) into a smaller data set with higher feature concentration. Use PDM AI engine to process it. The PDM AI engine has been trained using more than 10,000 models including normal and abnormal motor data.

Abnormal motor data type 1-high winding current

This type of data set represents early motor failures caused by overheated or burned motor windings. Normally, due to manufacturing tolerances or motor drive failure, one winding will fail before the other two windings. Connecting resistors in series with two “normal” windings can simulate this failure mode well. Figure 5 shows the “short circuit” in the simulated winding A. Figure 6 shows the folded image and the original image produced in this case. The circle has been deformed into an ellipse with its major axis on the x-axis. Table 1 summarizes the deformation caused by the high current of the three motor windings.

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 5. High current of motor winding A

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 6. The PDM image of the motor when the winding A current is high (Bad_Robot_D.jpg)

Table 1. Summary of PDM images when the motor windings are short-circuited

High current winding

Corresponding Iα C Iβ coordinate diagram

A

Ellipse with major axis on X axis

B

Ellipse with major axis at 45°

C

Ellipse with major axis at 135°

Abnormal motor data type 2-winding current is low

There are several situations that can cause a single winding current to be too low. For example, the connection part in a high-power motor may be corroded or loosened, causing the IR voltage to drop before the current reaches the motor windings. In addition, two of the windings may fail before the third winding, or the motor drive may become weaker in one of the phases. Similarly, we can simulate this failure by connecting a resistor in series with an “abnormal” motor winding, as shown in Figure 7. Figure 8 shows that the circle is deformed and becomes an ellipse with a major axis at 135 degrees. Table 2 summarizes the deformation caused by the low current of the three motor windings.

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 7. The impedance of motor winding B increases

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 8. PDM image of the motor with low winding B current (Bad_Robot_B.jpg)

Table 2. Summary of PDM images under low current of motor windings

Low current winding

Corresponding Iα C Iβ coordinate diagram

A

Ellipse with major axis at 90°

B

Ellipse with major axis at 135°

C

Ellipse with major axis at 45°

Abnormal motor data type 3-unbalanced load

The third type of fault also uses MCSA to detect the imbalance of the mechanical load on the motor. When the load is unbalanced, the moment of inertia is uneven and swings around the rotor axis (similar to the swing of the top before falling to the ground). When the moment of inertia swings around the motor shaft, the windings consume more or less current that is synchronized with the swing rather than the motor speed. In order to simulate this situation, you can fix an unbalanced inertia wheel on the motor shaft, and collect data after the motor reaches the operating speed. Figure 9 shows the PDM image of a motor with an unbalanced load. Problems with power management can also cause the same type of image.

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 9. PDM image of a motor with an unbalanced inertia wheel installed

Normal motor data-load balance current balance

So what are the operating characteristics of a normal motor? Figure 10 shows a PDM image of a motor with a balanced flywheel without any resistance connected. There are some slight deviations from the perfect circle in the directions of 60°, 170° and 290° (approximately 120° apart). This is the result of SV_PWM switching from one phase to the next phase.

Use the Lattice Automate solution set to achieve motor control with predictive maintenance capabilities
Figure 10. Motor PDM image when current and load are balanced

in conclusion

The predictive maintenance function for industrial motor control systems greatly reduces operating costs by minimizing system downtime caused by unexpected failures. The Lattice Automate solution has the hardware and software tools needed to quickly and easily implement PDM, and uses industry-standard MCSA solutions to escort the BLDC motors commonly used in many industrial applications (including robotics).

The Links:   LQ121S1DG11 CM75DUM-12F

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *