ZYNQ Integrated Vibration and Temperature Monitoring Terminal for Motor Fault Diagnosis, Targeting Online Monitoring of Industrial Motors, Gearboxes, Fans, Pumps, and Other Rotating Equipment

ZYNQ Integrated Vibration and Temperature Monitoring Terminal for Motor Fault Diagnosis

Electromechanical Equipment

Abstract: Addressing the online monitoring needs of industrial motors, gearboxes, fans, pumps, and other rotating equipment, this paper designs an integrated monitoring terminal based on the Xilinx ZYNQ heterogeneous architecture. The terminal integrates IEPE vibration acquisition and multi-channel temperature acquisition, utilizing FPGA (PL) to achieve highly synchronized, high-sampling-rate data acquisition, while ARM (PS) runs embedded Linux to perform feature extraction, fault diagnosis, and data upload. It can real-time capture typical defects such as motor unbalance, misalignment, bearing wear, and stator faults, making it suitable for industrial scenarios like smart factories and predictive maintenance (PdM).

I. Project Background

As core power equipment in industrial production, motors' operational status directly impacts production line stability. Traditional manual inspections are highly lagging and prone to missed detections; dedicated monitoring equipment, on the other hand, is costly, complex to deploy, and difficult to scale.

Typical faults in rotating machinery are manifested in:

Vibration Signals: Abnormal amplitude, spectrum, kurtosis, and crest factor.
Temperature Signals: Abnormal bearing temperature, casing temperature, and winding temperature.

Therefore, vibration + temperature fusion monitoring is the most mature and reliable technical route for motor fault diagnosis.

This project designs an integrated, low-power, highly reliable online monitoring terminal based on the ZYNQ-7000 series platform, achieving:

Multi-channel IEPE vibration synchronous acquisition
Multi-channel PT100/NTC temperature acquisition
Hardware-level real-time data buffering
Embedded AI / Threshold-based diagnosis
Ethernet / 4G upload to platform

II. System Overall Architecture

The system adopts the classic ZYNQ PL+PS heterogeneous architecture:

Engineering and Technology

1. Hardware Layer

IEPE Vibration Sensors × 1~4 channels
Temperature Sensors (PT100) × 4~8 channels
Signal Conditioning Circuitry (constant current source, amplification, anti-aliasing filtering)
High-speed ADC, analog switches, excitation circuit
ZYNQ7020/7010 Core Board
Ethernet, power supply, isolation protection

2. FPGA (PL) is responsible for:

1MHz~10kHz configurable sampling rate
Multi-channel synchronous sampling control
FIFO ping-pong buffering to prevent data loss
Digital filtering and downsampling
Temperature acquisition timing control

3. ARM (PS) is responsible for:

Data reading and parsing under Linux system
Time-domain / Frequency-domain feature calculation (RMS, peak-to-peak, kurtosis, spectrum)
Bearing fault characteristic frequency (BPFO, BPFI, BSF, FTF) calculation
Temperature threshold judgment and over-temperature alarm
Fault level determination and local storage
Modbus TCP/MQTT upload to cloud

4. Data Flow

Sensor → Conditioning → ADC → PL Acquisition Buffer → AXI DMA → PS Processing → Diagnosis → Upload / Local Storage

III. Key Hardware Design

1. IEPE Vibration Signal Conditioning

IEPE sensors require a 2~4mA constant current source for power, so the hardware must include:

Constant current source circuit (TL431 + op-amp or dedicated constant current chip)
AC coupling DC-blocking circuit
Low-pass anti-aliasing filter (cutoff frequency set according to sampling rate)
Level shifting circuit to adapt to ADC unipolar input

2. Temperature Acquisition Circuit

Utilizing:

PT100 platinum resistance temperature measurement
Constant current excitation + instrumentation amplifier
Multi-channel analog switch polling acquisition
Digital isolation to avoid strong electrical interference from motors

3. Grounding and Anti-Interference Design (Industrial Focus)

Single-point connection for analog and digital grounds
Power supply with added ferrite beads, LC filtering
TVS, RC absorption at signal input
Ethernet port magnetic isolation

Ensuring stable operation in strong industrial electromagnetic environments.

IV. FPGA (PL) Logic Design

1. Sampling Control Module

Built-in PLL generates precise sampling clock
Supports multiple configurations like 12.5k/25k/50k/100k
Multi-channel synchronous triggering to ensure time alignment of vibration and temperature

2. Dual Ping-Pong FIFO Buffer

With large data volumes during high-speed continuous acquisition, a ping-pong FIFO is used:

When FIFO A is written, FIFO B is read
Alternating switching, no data loss, no blocking
Suitable for long-term continuous acquisition

3. Simple Pre-processing (PL-side Acceleration)

Mean filtering
Truncation and data alignment
Oversampling averaging for noise reduction

Reduces ARM processing load and improves system real-time performance.

V. ARM (PS) Software and Fault Diagnosis Algorithms

The PS runs Linux, with the application layer implementing:

1. Data Reading

Read from PL-side FIFO via AXI DMA or AXI GP interface:

Raw vibration waveforms
Multi-channel temperature values
Timestamp information

2. Time-Domain Feature Calculation

Root Mean Square (RMS)
Peak-to-peak value
Kurtosis (sensitive to early faults)
Impulse factor, crest factor

3. Frequency-Domain Analysis (FFT)

Perform FFT on vibration signals to extract:

Rotational frequency
2x, 3x rotational frequency components (unbalance / misalignment)
Bearing fault characteristic frequencies
Harmonics and modulation phenomena

4. Fusion Diagnosis Logic

Vibration exceeding threshold → Pre-warning
Continuous temperature rise → Pre-warning
Vibration + temperature simultaneously abnormal → Severe alarm
Sudden change in kurtosis → Early impact-type faults (e.g., bearing pitting)

5. Upload and Storage

MQTT reporting to cloud platform
Local SD card black box storage
Supports real-time waveform viewing via web / PC software

VI. Typical Motor Fault Identification Capabilities

This terminal can effectively identify:

Rotor Unbalance Significant increase in 1x rotational frequency vibration
Coupling Misalignment Pronounced 2x, 3x rotational frequency components
Bearing Wear / Pitting Increased kurtosis, appearance of characteristic frequencies like BPFO/BPFI
Looseness (Base / End Cover) Increased low-frequency amplitude, chaotic spectrum
Stator / Electrical Faults Appearance of power frequency harmonic components in vibration
Overheating Risk Continuous rise in temperature curve

VII. System Advantages

ZYNQ Heterogeneous Advantage PL handles high-speed acquisition, PS handles intelligent diagnosis, achieving high efficiency through hardware-software co-design.
Strong Synchronicity Vibration and temperature data come with unified timestamps, facilitating correlation analysis.
Industrial-Grade Reliability Isolation, anti-interference, fanless, wide-temperature design.
Easy Deployment Standard Ethernet interface, compatible with MES, SCADA, and equipment cloud platforms.
Controllable Cost Over 50% cost reduction compared to imported equipment, suitable for large-scale deployment.

VIII. Application Scenarios

Online monitoring of high-voltage motors, asynchronous motors
Group control of fans, water pumps, air compressors
Fault pre-warning for gearboxes, reducers
Predictive maintenance for rotating equipment in coal mines, metallurgy, chemical industries
Data acquisition terminal for smart factory equipment digital twins