How to Design a Dual-Sensor Redundant Comparison System?

A dual-sensor redundant comparison system's fundamental design concept is "layered architecture + fault-tolerant logic." The entire procedure is standardized and may be carried out in four layers, from software algorithms to hardware selection:

 

1. Hardware Layer: Installation Design + Matching Selection

Ensuring consistency and minimizing variances from the start is fundamental to hardware design:

Sensor Selection: To guaranty initial characteristic matching, sensors must be from the same manufacturer, model, and manufacturing batch and have comparable nominal accuracy (e.g., both 0.05%FS);

Installation Layout: To guaranty entirely consistent operating circumstances, the two sensors should be placed side by side with a separation of ≤10mm; mounting hole coaxiality of ≤0.05mm and flatness of ≤0.02mm will guaranty consistent machining accuracy;

Acquisition Loop: Give sharing the same acquisition channel top priority. To guaranty constant channel faults, choose acquisition modules with the same specification if using different channels. Set aside interchangeable wiring interfaces for future troubleshooting.

 

2. Data Preprocessing Layer: Ensuring Signal Stability by Filtering Interference

To prevent errors brought on by interference, the raw signal is treated first:

Filtering: Moving average filtering is used to smooth the signal and filter electrical interference and pressure variations after median filtering has eliminated momentary spikes;

Outlier Removal: In the comparison calculation, points that are more than three times the historical fluctuation in a single period are eliminated;

Temperature Compensation: To combat zero-point drift brought on by temperature, synchronous temperature compensation is applied to the output after the sensor's internal temperature is measured.

 

3. Core Comparison Logic Layer: Fault-Tolerant Alarm + Dynamic Judgment

This is the system's central component, and the false alarm rate is determined by its design philosophy:

Stage-based Dynamic Thresholds: Avoids set thresholds by interacting with the equipment's PLC to obtain process information and automatically switching thresholds for various stages (e.g., injection molding ×2.0, cooling ×0.6);

Deviation Judgment Logic: Alarms are only activated when two requirements are satisfied: ① Deviation above the current threshold; ② Deviation exceeds the limit for ≥10 seconds, ignoring brief interference;

Graded Early Warning Mechanism: Continuously exceeding the limit and a single sensor output exceeding its accuracy range triggers a fault shutdown, preventing minor deviations from causing direct shutdowns and affecting production capacity. Approaching the threshold triggers a warning (recording only without shutting down); exceeding the limit triggers a secondary alarm.

 

4. Self-Maintenance Layer: Dynamic Adjustment + Automatic Calibration

minimizes manual maintenance while achieving long-term stable operation:

Automatic Zero-Point Alignment: Removes little drift from the previous day by automatically aligning the two sensors' zero points under zero-voltage steady-state circumstances upon starting;

Automatic Threshold Update: In order to account for the slow aging of sensors, thresholds are automatically updated using the 3σ principle based on weekly statistical analysis of historical deviation data;

Fault Location Function: When deviation surpasses thresholds, built-in exchange verification logic instantly initiates troubleshooting instructions, rapidly identifying whether the issue is with the sensor or the installation.

No sophisticated hardware is needed for the entire architecture; current PLCs can be utilized. Upgrades just need more expensive sensors for small and medium-sized companies; software logic can be set up using templates, requiring little development work.