Patent Assistant

Help me write a patent disclosure document:
I designed a system that uses an on-device model to detect abnormal vibration patterns in industrial motors.
It compresses sensor streams, runs anomaly detection locally, and only uploads event snippets to reduce bandwidth.

# Patent Technical Disclosure Document

## I. Title of Invention
A method and system for on-device vibration anomaly detection and event-based uploading for industrial motors

## II. Technical Field
The present invention relates to the technical field of industrial equipment monitoring, and specifically relates to on-device vibration signal processing and anomaly detection.

## III. Background Art
### 3.1 Description of Existing Technology
Existing solutions typically stream high-frequency vibration data to a cloud platform for centralized analysis, or use threshold-based alarms on edge devices.

### 3.2 Defects of Existing Technology
The existing technology has the following problems:
1. High bandwidth and storage costs due to continuous raw data uploading.
2. High latency for cloud-based detection, which may delay fault response.
3. Threshold-based edge alarms have poor adaptability across different motor types and operating conditions.

## IV. Summary of Invention
### 4.1 Technical Problem to be Solved
The technical problem to be solved by the present invention is reducing bandwidth and latency while maintaining reliable anomaly detection for motor vibration monitoring.

### 4.2 Technical Solution
To solve the above technical problem, the present invention adopts the following technical solution:
- Collect vibration sensor data from an industrial motor using an edge device.
- Perform stream compression/feature extraction on the edge device to obtain compact representations.
- Execute an on-device anomaly detection model to identify abnormal vibration patterns.
- When an anomaly is detected, generate an event snippet containing a time window of raw/feature data and metadata (timestamp, motor ID, operating state).
- Upload only the event snippet (and optionally periodic summaries) to a remote server for storage, visualization, and further diagnosis.

### 4.3 Beneficial Effects
By adopting the technical solution of the present invention, the following beneficial effects are achieved:
1. Significantly reduced network bandwidth usage by avoiding continuous raw data uploads.
2. Faster anomaly response due to local inference and event-triggered reporting.
3. Improved detection robustness compared with fixed thresholds by using a learned model.

## V. Detailed Description
### 5.1 Embodiment 1
An edge device connected to an accelerometer samples vibration at a preset rate, computes spectral features, and runs an anomaly model. Upon detection, it uploads a 5-second window around the event plus operating metadata.

### 5.2 Embodiment 2 (Optional)
The anomaly model is periodically updated using federated or scheduled offline training, while inference remains on-device.

## VI. Description of Drawings
Suggestions for drawing the following figures:
- Figure 1: System architecture (sensor, edge device, network, server).
- Figure 2: On-device processing pipeline (sampling → compression/features → anomaly detection → event packaging → upload).

## VII. Keywords
vibration monitoring; anomaly detection; edge computing; event-based upload; industrial motor; signal compression

python scripts/patent_search.py "vibration anomaly detection edge event-based upload" --limit 20

python scripts/patent_search.py "vibration anomaly detection edge event-based upload" -s all -p

python scripts/patent_search.py "vibration anomaly detection edge event-based upload" -s google,cnipa,innojoy

python scripts/patent_search.py "vibration anomaly detection edge event-based upload" -s all -p -a