MBG0903 commited on
Commit
2fb45f0
Β·
verified Β·
1 Parent(s): 27dd7da

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +141 -12
README.md CHANGED
@@ -1,20 +1,149 @@
1
  ---
2
- title: Machine Prediction Model AIEMS
3
- emoji: πŸš€
4
- colorFrom: red
5
- colorTo: red
6
  sdk: docker
7
- app_port: 8501
8
- tags:
9
- - streamlit
10
  pinned: false
11
- short_description: Machine_Prediction_Model_AIEMS
12
  license: mit
13
  ---
14
 
15
- # Welcome to Streamlit!
16
 
17
- Edit `/src/streamlit_app.py` to customize this app to your heart's desire. :heart:
18
 
19
- If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
20
- forums](https://discuss.streamlit.io).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ title: Intelligent Predictive Maintenance System
3
+ emoji: 🏭
4
+ colorFrom: blue
5
+ colorTo: indigo
6
  sdk: docker
7
+ app_port: 7860
 
 
8
  pinned: false
 
9
  license: mit
10
  ---
11
 
12
+ # Intelligent Predictive Maintenance System
13
 
14
+ A Streamlit-based educational prototype demonstrating collaboration between a trained Machine Learning model and Agentic AI components.
15
 
16
+ ## Architecture
17
+
18
+ ```text
19
+ IoT Machine Parameters
20
+ ↓
21
+ IoT Data Agent
22
+ ↓
23
+ ML Prediction Agent
24
+ ↓
25
+ Trained Random Forest Pipeline
26
+ ↓
27
+ Failure Probability + Risk Level
28
+ ↓
29
+ Engineering Reasoning Agent
30
+ ↓
31
+ Qwen/Qwen2.5-0.5B-Instruct
32
+ ↓
33
+ Maintenance Action Agent
34
+ ↓
35
+ Structured Work Order
36
+ ↓
37
+ Human Engineer Approval
38
+ ```
39
+
40
+ ## Machine Learning Model
41
+
42
+ The application uses the pre-trained file:
43
+
44
+ ```text
45
+ predictive_maintenance_rf_pipeline.pkl
46
+ ```
47
+
48
+ The pipeline was trained separately using the AI4I 2020 predictive-maintenance dataset.
49
+
50
+ The expected raw input features are:
51
+
52
+ - Type
53
+ - Air temperature [K]
54
+ - Process temperature [K]
55
+ - Rotational speed [rpm]
56
+ - Torque [Nm]
57
+ - Tool wear [min]
58
+
59
+ The application does **not** retrain the Random Forest during startup. The exported pipeline is loaded and used only for inference.
60
+
61
+ ## Local Reasoning Model
62
+
63
+ The engineering reasoning layer uses:
64
+
65
+ ```text
66
+ Qwen/Qwen2.5-0.5B-Instruct
67
+ ```
68
+
69
+ The model receives:
70
+
71
+ - Current sensor readings
72
+ - Random Forest failure probability
73
+ - Random Forest prediction
74
+ - Calculated risk level
75
+
76
+ The language model explains the prediction and proposes an educational maintenance response. It does not replace the Random Forest prediction.
77
+
78
+ ## Agent Design
79
+
80
+ ### Agent 1 β€” IoT Data Agent
81
+
82
+ Validates and structures incoming sensor data.
83
+
84
+ ### Agent 2 β€” ML Prediction Agent
85
+
86
+ Calls the trained Random Forest pipeline as a prediction tool.
87
+
88
+ ### Agent 3 β€” Engineering Reasoning Agent
89
+
90
+ Uses Qwen to interpret the ML result and sensor readings.
91
+
92
+ ### Agent 4 β€” Maintenance Action Agent
93
+
94
+ Creates a structured predictive-maintenance work order.
95
+
96
+ ### Human-in-the-Loop
97
+
98
+ A qualified engineer must validate any recommendation before a real maintenance, repair, or shutdown decision.
99
+
100
+ ## Hugging Face Space Deployment
101
+
102
+ Create a new Hugging Face Space and choose:
103
+
104
+ ```text
105
+ SDK: Docker
106
+ ```
107
+
108
+ Upload these files to the root of the Space:
109
+
110
+ ```text
111
+ app.py
112
+ requirements.txt
113
+ Dockerfile
114
+ README.md
115
+ predictive_maintenance_rf_pipeline.pkl
116
+ ```
117
+
118
+ The final repository should look like:
119
+
120
+ ```text
121
+ /
122
+ β”œβ”€β”€ app.py
123
+ β”œβ”€β”€ requirements.txt
124
+ β”œβ”€β”€ Dockerfile
125
+ β”œβ”€β”€ README.md
126
+ └── predictive_maintenance_rf_pipeline.pkl
127
+ ```
128
+
129
+ The Docker container exposes Streamlit on port `7860`, which matches the Hugging Face Space configuration.
130
+
131
+ ## Hardware
132
+
133
+ The Random Forest model is lightweight.
134
+
135
+ The local Qwen 0.5B model can run on CPU, but response generation may take several seconds depending on the Space hardware. CPU Upgrade or GPU hardware will provide a smoother live demonstration.
136
+
137
+ ## Important Model Compatibility Note
138
+
139
+ The exported Random Forest pipeline was created with scikit-learn 1.6.1, so this project pins:
140
+
141
+ ```text
142
+ scikit-learn==1.6.1
143
+ ```
144
+
145
+ Keeping the same scikit-learn version is important when loading a Joblib/Pickle model.
146
+
147
+ ## Educational Disclaimer
148
+
149
+ This application is intended only for teaching and demonstration. It must not be used to make real industrial maintenance, repair, shutdown, or safety decisions without validation by qualified engineering personnel.