Abstract
The vibration measuring standard for compound machinery utilized in modern industrial production will be employed for the application of detecting technologies. The vibration intensity can be obtained by selecting the detecting method to obtain the speed of mechanical vibration, and technicians can examine whether the vibrating machinery is in a proper functioning state based on the value of vibration intensity, allowing for thorough fault diagnosis. In order to provide useful diagnosis ideas for technicians, this study examines the measurement of mechanical vibration and investigates the calculating method of mechanical vibration intensity.
Introduction
In contemporary industrial production, pumps are indispensable for transporting a wide variety of fluids, from water and oil to complex suspended solids. Many of these units are sophisticated, often imported machines, which can present significant diagnostic challenges for local maintenance teams. Traditional maintenance often relies on subjective observations—such as auditory cues—which lack the precision required for modern reliability standards.
The primary objective of this research is to bridge the gap between subjective observation and objective, data-driven diagnostics. By establishing a standardized framework for measuring and calculating vibration intensity, technicians can accurately assess whether a pump is operating within safe parameters or requires immediate intervention to prevent catastrophic failure.
Methodology
The proposed diagnostic workflow integrates hardware sensing with advanced digital signal processing (DSP). The process begins with the strategic placement of piezoelectric acceleration sensors at critical mechanical junctions, specifically the pump bearings and hydraulic ends.
The data acquisition chain involves several key steps:
- Signal Acquisition: Acceleration signals are captured and amplified via a KD5018 charge amplifier before being digitized through an NI9205 data acquisition card.
- Signal Processing: Using MATLAB, digital frequency domain filtering is applied. A vital component of this stage is polynomial fitting, used to eliminate "trend terms" or low-frequency drift, ensuring the integrity of the resulting velocity signal.
- Intensity Calculation: Vibration intensity is quantified as the Root Mean Square (RMS) value of the vibration velocity within the 10–1000 Hz frequency range.
- Standardized Rating: The calculated values are benchmarked against the "Method for Measuring and Rating Vibration of Reciprocating Machines" standard. This standard classifies machinery into four classes (I-IV) and assigns quality grades: Excellent, Good, Allowed, and Not Allowed.
Results and Discussion
The experimental phase utilized an F-1300 ordinary drilling pump as the primary test subject. Data was collected across three axes (X, Y, and Z) at multiple test points to capture a comprehensive vibration profile.
Key findings from the analysis include:
- Directional Sensitivity: Vibration intensity was found to be non-uniform across different axes. At the pump bearing, the X-direction showed the highest intensity (mean of 3.1928 mm/s), while the hydraulic end was most active in the Y-direction (mean of 1.2771 mm/s). This underscores the necessity of multi-axis monitoring for accurate diagnosis.
- Health Classification: For the F-1300 pump (categorized as Class IV large machinery), the measured intensities translated to "Good" and "Excellent" ratings, respectively. This provides a clear, standardized "go/no-go" signal for operators.
- Frequency Thresholds: Normal operation for this specific pump model typically remains within a 5000 Hz threshold. Sampling frequencies significantly exceeding this limit serve as a primary indicator of mechanical abnormality.
Conclusion
This study demonstrates that standardized vibration intensity measurements provide a robust and efficient means of diagnosing pump health. The MATLAB-based frequency domain filtering approach is computationally efficient and avoids time-shift artifacts, making it highly suitable for real-world industrial applications.
While the current method excels at identifying overall health status, future research should focus on developing granular spectral "fingerprints" to isolate specific component failures, such as distinguishing between worn bearings and failing valves. Integrating these methods into automated, real-time monitoring systems will further enhance industrial reliability and safety.
Summary updated