Multimodal Data Fusion Analysis Based on Excel: Intelligent Sensor Network Decision System

In the fields of industrial IoT and smart manufacturing, efficiently processing and fusing heterogeneous data from multiple sensors through Excel to achieve real-time monitoring and intelligent decision-making is an important technical challenge. This article will introduce how to utilize Excel to build a multimodal data fusion analysis platform for the collection, processing, and decision support of multisource sensor data such as temperature, pressure, and vibration.

Practical Application Scenario:

The equipment management department of a smart factory needs to monitor the operational status of multiple critical devices in real-time. Each device is equipped with temperature sensors (°C), pressure sensors (MPa), and vibration sensors (mm/s), with data collected every minute. Managers need to construct an Excel-based analysis platform to assess equipment health status and warn of potential failures by fusing these heterogeneous data.

Operation Steps:

1. Data Import and Preprocessing

– Create a new workbook and set three worksheets named “Temperature Data”, “Pressure Data”, and “Vibration Data” respectively.

– Use Power Query (shortcut: ALT+N+PF) to connect to the sensor data source.

– Create a unified data structure in each worksheet:

Column A: Timestamp

Column B: Device ID

Column C: Sensor Value

Column D: Data Quality Indicator

2. Data Synchronization and Alignment

Create a new worksheet “Fused Data”, and enter the formula in cell A1:

=VLOOKUP(Timestamp, Temperature Data!A:C, 3, FALSE)

Similarly, add lookup formulas for pressure and vibration data.

3. Implement Data Fusion Evaluation System

Use the following VBA code to create the data fusion analysis module:

“`vba

Public Sub SensorDataFusion()

Dim ws As Worksheet

Dim lastRow As Long

Dim i As Long

Dim healthScore As Double

‘ Initialize weight coefficients

Const TEMP_WEIGHT As Double = 0.4

Const PRESS_WEIGHT As Double = 0.3

Const VIB_WEIGHT As Double = 0.3

‘ Set thresholds

Const TEMP_THRESHOLD As Double = 85

Const PRESS_THRESHOLD As Double = 12

Const VIB_THRESHOLD As Double = 25

Set ws = ThisWorkbook.Sheets(“Fused Data”)

lastRow = ws.Cells(ws.Rows.Count, “A”).End(xlUp).Row

‘ Add health status evaluation column

ws.Range(“E1”).Value = “Device Health Score”

ws.Range(“F1”).Value = “Status Evaluation”

‘ Implement Dempster-Shafer evidence theory fusion

For i = 2 To lastRow

‘ Calculate basic probability assignment for each sensor

Dim tempBPA As Double

Dim pressBPA As Double

Dim vibBPA As Double

tempBPA = Calculate_BPA(ws.Cells(i, “B”).Value, TEMP_THRESHOLD)

pressBPA = Calculate_BPA(ws.Cells(i, “C”).Value, PRESS_THRESHOLD)

vibBPA = Calculate_BPA(ws.Cells(i, “D”).Value, VIB_THRESHOLD)

‘ Fusion calculation

healthScore = DS_Fusion(tempBPA, pressBPA, vibBPA, _

TEMP_WEIGHT, PRESS_WEIGHT, VIB_WEIGHT)

‘ Record results

ws.Cells(i, “E”).Value = healthScore

ws.Cells(i, “F”).Value = Evaluate_Health(healthScore)

Next i

‘ Create status monitoring chart

CreateHealthMonitorChart ws

End Sub

Private Function Calculate_BPA(value As Double, threshold As Double) As Double

‘ Basic probability assignment calculation based on Gaussian distribution

Calculate_BPA = Exp(-((value – threshold) ^ 2) / (2 * threshold ^ 2))

End Function

Private Function DS_Fusion(temp As Double, press As Double, vib As Double, _

w1 As Double, w2 As Double, w3 As Double) As Double

‘ D-S evidence theory fusion algorithm implementation

Dim k As Double ‘ Conflict coefficient

k = 1 – (temp * press * vib)

DS_Fusion = (w1 * temp + w2 * press + w3 * vib) / (1 + k)

End Function

Private Function Evaluate_Health(score As Double) As String

‘ Health status evaluation

Select Case score

Case Is >= 0.9

Evaluate_Health = “Normal”

Case 0.7 To 0.9

Evaluate_Health = “Caution”

Case Else

Evaluate_Health = “Warning”

End Select

End Function

“`

Code Explanation:

– Calculate_BPA function: Calculates the basic probability assignment for each sensor’s data.

– DS_Fusion function: Implements data fusion based on D-S evidence theory.

– Evaluate_Health function: Assesses the health status of the device based on the fusion score.

– TEMP_WEIGHT and other constants: Define the weight coefficients for each sensor’s data.

4. Add Real-Time Monitoring Chart

Create a dynamic monitoring chart in the “Fused Data” worksheet:

– Select data range A1:F[lastRow]

– Insert a line chart (shortcut: ALT+N+L)

– Set primary and secondary axes to display sensor data with different dimensions

– Add threshold reference lines for health scores

Tips and Precautions:

– It is recommended to enable Excel’s “Data Analysis Toolpak” to support advanced statistical analysis.

– For large volumes of real-time data, consider using Power Pivot for data management.

– An appropriate error handling mechanism is needed when sensor data is abnormal.

– Regularly calibrate weight coefficients to optimize fusion results.

– Consider adding data smoothing to reduce noise impact.

– It is advisable to regularly back up data and analysis results.

– Monitor Excel’s memory usage during system operation.

The multimodal data fusion analysis system introduced in this article can effectively integrate multisource sensor data, providing decision support for equipment status monitoring and predictive maintenance. The intelligent fusion algorithm implemented through D-S evidence theory can comprehensively consider the reliability and uncertainty of various sensor data, providing more accurate equipment health assessment results. This system is suitable for small to medium-sized industrial monitoring scenarios, but when handling high-frequency large-capacity data, it may be necessary to consider using professional database systems.

Leave a Comment