Bridge structural-health anomaly detection
A structural-health analysis laboratory for detecting changes in prepared bridge vibration, strain, displacement, and environmental sensor data.
Project definition
Problem statement
Bridge sensors respond to traffic, temperature, wind, measurement faults, and structural condition. An unusual signal does not automatically mean structural damage.
The engineering problem is to establish a traceable baseline, remove or model environmental effects, detect meaningful changes, and show evidence and uncertainty for engineer review.
Project objectives
- Prepare synchronised vibration, strain, displacement, temperature, and event data for a selected bridge model.
- Detect missing samples, clipping, drift, time misalignment, and sensor failure.
- Extract frequency, modal, statistical, and trend features from defined analysis windows.
- Compare threshold, statistical, and selected machine-learning anomaly methods.
- Measure detection performance across prepared damage, traffic, weather, and sensor-fault scenarios.
System design
System modules
Sensor registry
Stores sensor type, location, orientation, sampling rate, calibration, units, and installation metadata.
Signal-quality processor
Checks timing, gaps, spikes, clipping, drift, noise, and cross-sensor consistency.
Feature engine
Calculates spectral peaks, modal indicators, damping proxies, strain statistics, trends, and environmental features.
Anomaly detector
Builds a normal baseline and scores changes using transparent and selected learned methods.
Engineering dashboard
Shows sensor layouts, signals, modes, trends, events, explanations, confidence, and review decisions.
Methodology
System workflow
- 01Load bridge model
The student selects prepared bridge, sensor, environmental, and baseline records.
- 02Validate signals
Quality issues are detected and affected windows are corrected, excluded, or marked.
- 03Build baseline
Normal data defines expected features across documented traffic and temperature ranges.
- 04Run scenarios
Prepared structural-change and sensor-fault cases are processed without revealing labels to the detector.
- 05Review anomalies
The dashboard links scores to sensors, features, conditions, and similar baseline periods.
Demonstration scenario
A prepared bridge dataset contains normal temperature and traffic variation followed by a small modal-frequency change. The system first rejects a faulty sensor window, then shows a persistent multi-sensor anomaly after environmental normalisation. The student reviews the affected modes and records an engineering-inspection recommendation.
Engineering
Technical architecture
- Web application
- Nuxt and Vue for sensor maps, signal charts, mode views, anomaly timelines, and review.
- Analysis API
- FastAPI for datasets, signal jobs, features, baselines, anomalies, labels, and exports.
- Data layer
- PostgreSQL for bridge and sensor metadata, window summaries, features, models, events, and reviews.
- Signal pipeline
- Python, NumPy, and SciPy for filtering, spectral analysis, modal features, synchronisation, and quality checks.
- Anomaly evaluation
- scikit-learn and statistical baselines with time-separated tests, environmental normalisation, and fault scenarios.
Testing
Evaluation
Evaluation measures
- Detection precision, recall, and delay across prepared anomaly scenarios
- False-alert rate under normal traffic and environmental variation
- Frequency, modal, and statistical feature error against reference signals
- Ability to distinguish sensor faults from prepared structural-change cases
- Robustness to missing data, noise, sensor loss, and temperature shift
- Processing time across sensor counts, sample rates, and monitoring windows
System boundaries
- The project is an educational analysis system and not a certified bridge safety monitor.
- An anomaly indicates a change requiring review and does not diagnose damage or structural safety.
- Demonstrations use simulated, public, or explicitly permitted sensor data.
- Inspection, load restriction, closure, or repair decisions belong to qualified bridge engineers and responsible authorities.
Included
- 01Bridge sensor and experiment-management interface
- 02Signal validation, feature, normalisation, and anomaly modules
- 03Mode, trend, event, confidence, and error-analysis dashboards
- 04Prepared simulated or public sensor data with results
- 05Complete source code in a private GitHub repository
- 06Complete project documentation with synopsis, abstract, methodology, bridge and sensor diagrams, experiment results, screenshots, and conclusion
- 07Setup and usage guide
Project record
No buyer information is collected on this page.
- Permanent project ID
- GP-CV-0ZGQ2Q1
- Catalogued
- 21 Aug 2026
- Completed
- Pending
- Verified
- Pending
- Demonstration
- Added when ready
Handover
After purchase
- 01Payment is confirmed
The project is marked unavailable and cannot be purchased again.
- 02Repository access is granted
The buyer's submitted GitHub account receives access to the private repository.
- 03The purchase record is delivered
The certification sheet is prepared from the reviewed buyer details and sent privately by email.