File size: 19,226 Bytes
0e1a9c0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 | import json
from datetime import datetime
from pathlib import Path
import joblib
import pandas as pd
import streamlit as st
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
# ============================================================
# APP CONFIGURATION
# ============================================================
st.set_page_config(
page_title="Intelligent Predictive Maintenance System",
page_icon="π",
layout="wide",
)
MODEL_PATH = Path(__file__).parent / "predictive_maintenance_rf_pipeline.pkl"
LLM_MODEL_ID = "Qwen/Qwen2.5-0.5B-Instruct"
FAILURE_THRESHOLD = 0.50
MEDIUM_RISK_THRESHOLD = 0.40
HIGH_RISK_THRESHOLD = 0.70
# ============================================================
# LIGHT UI STYLING
# ============================================================
st.markdown(
"""
<style>
.main-title {
font-size: 2.2rem;
font-weight: 750;
margin-bottom: 0.1rem;
}
.sub-title {
color: #6b7280;
margin-bottom: 1.25rem;
}
.agent-box {
border: 1px solid #d1d5db;
border-radius: 12px;
padding: 14px 16px;
margin-bottom: 12px;
}
.status-ok {
font-weight: 700;
}
.small-note {
font-size: 0.88rem;
color: #6b7280;
}
</style>
""",
unsafe_allow_html=True,
)
# ============================================================
# LOAD TRAINED RANDOM FOREST PIPELINE
# ============================================================
@st.cache_resource
def load_predictive_model():
return joblib.load(MODEL_PATH)
# ============================================================
# LOAD LOCAL QWEN MODEL
# ============================================================
@st.cache_resource
def load_reasoning_model():
device = "cuda" if torch.cuda.is_available() else "cpu"
tokenizer = AutoTokenizer.from_pretrained(
LLM_MODEL_ID
)
model = AutoModelForCausalLM.from_pretrained(
LLM_MODEL_ID,
torch_dtype=torch.float16 if device == "cuda" else torch.float32,
low_cpu_mem_usage=True,
)
model.to(device)
model.eval()
return tokenizer, model, device
# ============================================================
# AGENT 1 β IoT DATA AGENT
# ============================================================
def iot_data_agent(
product_type,
air_temperature,
process_temperature,
rotational_speed,
torque,
tool_wear,
):
"""
Validates and packages incoming IoT-style machine readings.
"""
errors = []
if product_type not in {"L", "M", "H"}:
errors.append("Product Type must be L, M or H.")
if air_temperature <= 0:
errors.append("Air temperature must be greater than 0 K.")
if process_temperature <= 0:
errors.append("Process temperature must be greater than 0 K.")
if rotational_speed <= 0:
errors.append("Rotational speed must be greater than 0 rpm.")
if torque < 0:
errors.append("Torque cannot be negative.")
if tool_wear < 0:
errors.append("Tool wear cannot be negative.")
if errors:
return None, errors
sensor_record = pd.DataFrame(
[
{
"Type": product_type,
"Air temperature [K]": float(air_temperature),
"Process temperature [K]": float(process_temperature),
"Rotational speed [rpm]": float(rotational_speed),
"Torque [Nm]": float(torque),
"Tool wear [min]": float(tool_wear),
}
]
)
return sensor_record, []
# ============================================================
# AGENT 2 β ML PREDICTION AGENT
# ============================================================
def prediction_agent(sensor_record):
"""
Calls the trained Random Forest pipeline as a prediction tool.
"""
model = load_predictive_model()
failure_probability = float(
model.predict_proba(sensor_record)[0, 1]
)
prediction = int(
failure_probability >= FAILURE_THRESHOLD
)
if failure_probability >= HIGH_RISK_THRESHOLD:
risk_level = "HIGH"
elif failure_probability >= MEDIUM_RISK_THRESHOLD:
risk_level = "MEDIUM"
else:
risk_level = "LOW"
return {
"prediction": prediction,
"failure_probability": failure_probability,
"risk_level": risk_level,
}
# ============================================================
# AGENT 3 β ENGINEERING REASONING AGENT
# ============================================================
def build_engineering_prompt(sensor_record, ml_result):
row = sensor_record.iloc[0]
system_prompt = """
You are an engineering maintenance reasoning assistant in an educational
predictive-maintenance prototype.
A trained Random Forest model has already produced the machine-failure
probability. Do not replace or contradict the ML prediction.
Use only the supplied sensor readings and ML result. Do not invent
manufacturer limits, maintenance history, measurements, failure codes,
or engineering standards that were not provided.
Give a concise engineering interpretation suitable for final-year
engineering students.
Return exactly these headings:
### Engineering Assessment
### Evidence from Current Readings
### Possible Engineering Concern
### Recommended Next Action
### Preventive Maintenance Suggestion
Always state that the recommendation requires validation by a qualified
engineer before any real maintenance, shutdown, or repair decision.
""".strip()
user_prompt = f"""
Machine readings:
Product Type: {row['Type']}
Air Temperature: {row['Air temperature [K]']:.1f} K
Process Temperature: {row['Process temperature [K]']:.1f} K
Rotational Speed: {row['Rotational speed [rpm]']:.0f} rpm
Torque: {row['Torque [Nm]']:.1f} Nm
Tool Wear: {row['Tool wear [min]']:.0f} min
Random Forest result:
Failure Probability: {ml_result['failure_probability']:.1%}
Risk Level: {ml_result['risk_level']}
Prediction: {'Failure Risk' if ml_result['prediction'] == 1 else 'Healthy'}
Explain what this result means from a predictive-maintenance perspective.
""".strip()
return [
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt},
]
def engineering_reasoning_agent(sensor_record, ml_result):
"""
Uses Qwen to interpret the ML output and sensor readings.
"""
tokenizer, model, device = load_reasoning_model()
messages = build_engineering_prompt(sensor_record, ml_result)
prompt_text = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True,
)
inputs = tokenizer(
prompt_text,
return_tensors="pt",
truncation=True,
max_length=1536,
).to(device)
with torch.inference_mode():
output_ids = model.generate(
**inputs,
max_new_tokens=220,
do_sample=False,
repetition_penalty=1.05,
pad_token_id=tokenizer.eos_token_id,
)
generated_tokens = output_ids[
0,
inputs["input_ids"].shape[1]:
]
response = tokenizer.decode(
generated_tokens,
skip_special_tokens=True,
).strip()
return response
# ============================================================
# AGENT 4 β MAINTENANCE ACTION AGENT
# ============================================================
def maintenance_agent(sensor_record, ml_result):
"""
Creates a structured action/work-order record from the ML result.
The engineering reasoning text is generated separately by Qwen.
"""
probability = ml_result["failure_probability"]
risk = ml_result["risk_level"]
if risk == "HIGH":
priority = "P1 - Immediate Engineering Review"
action = "Inspect the equipment before continued operation."
elif risk == "MEDIUM":
priority = "P2 - Priority Inspection"
action = "Schedule a maintenance inspection and increase monitoring."
else:
priority = "P3 - Routine Monitoring"
action = "Continue monitoring under the normal maintenance plan."
work_order_id = datetime.now().strftime("PM-%Y%m%d-%H%M%S")
return {
"work_order_id": work_order_id,
"work_order_type": "AI-Assisted Predictive Maintenance",
"priority": priority,
"failure_probability": f"{probability:.1%}",
"risk_level": risk,
"recommended_action": action,
"assigned_team": "Reliability / Maintenance Engineering",
"approval_status": "Pending Human Engineer Approval",
}
# ============================================================
# ORCHESTRATOR
# ============================================================
def run_agentic_workflow(
product_type,
air_temperature,
process_temperature,
rotational_speed,
torque,
tool_wear,
):
sensor_record, errors = iot_data_agent(
product_type,
air_temperature,
process_temperature,
rotational_speed,
torque,
tool_wear,
)
if errors:
return {
"success": False,
"errors": errors,
}
ml_result = prediction_agent(sensor_record)
reasoning = engineering_reasoning_agent(
sensor_record,
ml_result,
)
work_order = maintenance_agent(
sensor_record,
ml_result,
)
return {
"success": True,
"sensor_record": sensor_record,
"ml_result": ml_result,
"reasoning": reasoning,
"work_order": work_order,
}
# ============================================================
# SIDEBAR
# ============================================================
with st.sidebar:
st.header("About the Prototype")
st.write(
"""
This application demonstrates how a trained Machine Learning
model can be used as a tool inside an Agentic AI workflow.
"""
)
st.markdown(
"""
**Architecture**
1. π‘ IoT Data Agent
2. π€ ML Prediction Agent
3. π§ Engineering Reasoning Agent
4. π§ Maintenance Action Agent
5. π¨βπ§ Human Approval
"""
)
st.caption(
"Educational prototype only. Not intended for real industrial safety decisions."
)
# ============================================================
# MAIN PAGE
# ============================================================
st.markdown(
'<div class="main-title">π Intelligent Predictive Maintenance System</div>',
unsafe_allow_html=True,
)
st.markdown(
'<div class="sub-title">Random Forest + Local Qwen Reasoning + Agentic AI Workflow</div>',
unsafe_allow_html=True,
)
st.info(
"Workflow: IoT Data β Random Forest Prediction β Qwen Engineering Reasoning β Maintenance Action β Human Approval"
)
input_col, result_col = st.columns([1, 1.25], gap="large")
# ============================================================
# INPUT PANEL
# ============================================================
with input_col:
st.subheader("π‘ IoT Machine Parameters")
product_type = st.selectbox(
"Product Type",
options=["L", "M", "H"],
index=0,
help="L = Low, M = Medium, H = High product quality type used in the AI4I dataset.",
)
air_temperature = st.slider(
"Air Temperature [K]",
min_value=295.0,
max_value=305.0,
value=300.0,
step=0.1,
)
process_temperature = st.slider(
"Process Temperature [K]",
min_value=305.0,
max_value=315.0,
value=310.0,
step=0.1,
)
rotational_speed = st.slider(
"Rotational Speed [rpm]",
min_value=1100,
max_value=2900,
value=1500,
step=10,
)
torque = st.slider(
"Torque [Nm]",
min_value=3.0,
max_value=80.0,
value=45.0,
step=0.5,
)
tool_wear = st.slider(
"Tool Wear [min]",
min_value=0,
max_value=253,
value=150,
step=5,
)
run_button = st.button(
"π Run Agentic AI Workflow",
type="primary",
use_container_width=True,
)
# ============================================================
# OUTPUT PANEL
# ============================================================
with result_col:
st.subheader("π€ Machine Intelligence")
if not run_button:
st.write(
"Enter the operating parameters and run the workflow to see the agents collaborate."
)
if run_button:
with st.status(
"Running collaborative AI workflow...",
expanded=True,
) as status:
st.write("π‘ **Agent 1 β IoT Data Agent:** Validating sensor readings...")
sensor_record, errors = iot_data_agent(
product_type,
air_temperature,
process_temperature,
rotational_speed,
torque,
tool_wear,
)
if errors:
for error in errors:
st.error(error)
status.update(
label="Input validation failed",
state="error",
)
st.stop()
st.write("β
Sensor validation complete.")
st.write("π€ **Agent 2 β ML Prediction Agent:** Calling trained Random Forest model...")
try:
ml_result = prediction_agent(sensor_record)
except Exception as exc:
st.error(f"Unable to load or run the trained ML model: {exc}")
st.info(
"Ensure predictive_maintenance_rf_pipeline.pkl is in the same folder as app.py "
"and that scikit-learn==1.6.1 is installed."
)
status.update(
label="ML prediction failed",
state="error",
)
st.stop()
st.write("β
Random Forest prediction complete.")
st.write("π§ **Agent 3 β Engineering Reasoning Agent:** Qwen is interpreting the result...")
try:
reasoning = engineering_reasoning_agent(
sensor_record,
ml_result,
)
except Exception as exc:
reasoning = (
"The local language model could not generate an engineering explanation. "
f"Technical message: {exc}"
)
st.write("β
Engineering interpretation complete.")
st.write("π§ **Agent 4 β Maintenance Action Agent:** Creating maintenance action...")
work_order = maintenance_agent(
sensor_record,
ml_result,
)
st.write("β
Maintenance work order generated.")
status.update(
label="Agentic AI workflow completed",
state="complete",
)
probability = ml_result["failure_probability"]
prediction_text = (
"FAILURE RISK"
if ml_result["prediction"] == 1
else "HEALTHY"
)
metric1, metric2, metric3 = st.columns(3)
metric1.metric(
"Machine Status",
prediction_text,
)
metric2.metric(
"Failure Probability",
f"{probability:.1%}",
)
metric3.metric(
"Risk Level",
ml_result["risk_level"],
)
st.progress(
min(max(probability, 0.0), 1.0),
text=f"Predicted failure probability: {probability:.1%}",
)
with st.expander(
"π‘ Agent 1 β IoT Data Agent",
expanded=False,
):
st.dataframe(
sensor_record,
use_container_width=True,
hide_index=True,
)
with st.expander(
"π€ Agent 2 β ML Prediction Agent",
expanded=True,
):
st.write(
"The trained Random Forest pipeline was called as an ML prediction tool."
)
st.json(
{
"prediction": prediction_text,
"failure_probability": round(probability, 4),
"risk_level": ml_result["risk_level"],
}
)
with st.expander(
"π§ Agent 3 β Engineering Reasoning Agent",
expanded=True,
):
st.markdown(reasoning)
st.caption(
f"Local reasoning model: {LLM_MODEL_ID}"
)
with st.expander(
"π§ Agent 4 β Maintenance Action Agent",
expanded=True,
):
st.json(work_order)
st.warning(
"π¨βπ§ Human-in-the-loop: A qualified engineer must validate the recommendation "
"before any real maintenance, repair, or shutdown action."
)
report = {
"sensor_readings": sensor_record.iloc[0].to_dict(),
"ml_prediction": {
"prediction": prediction_text,
"failure_probability": probability,
"risk_level": ml_result["risk_level"],
},
"engineering_reasoning": reasoning,
"maintenance_work_order": work_order,
}
st.download_button(
"β¬οΈ Download AI Maintenance Report",
data=json.dumps(report, indent=2),
file_name="predictive_maintenance_report.json",
mime="application/json",
use_container_width=True,
)
# ============================================================
# EXPLANATION TAB AREA
# ============================================================
st.divider()
with st.expander("π How this Agentic AI system works"):
st.markdown(
"""
### 1. Random Forest is the prediction intelligence
The model was trained separately using the AI4I predictive-maintenance dataset.
It calculates the probability of machine failure from the current machine readings.
### 2. The Prediction Agent uses the RF model as a tool
The agent does not ask the language model to guess whether the machine will fail.
### 3. Qwen provides engineering interpretation
The local instruction-tuned model receives both the sensor readings and the
Random Forest prediction and explains the result in natural language.
### 4. The Maintenance Agent converts intelligence into action
It creates a structured maintenance work-order record.
### 5. Human approval remains mandatory
Agentic AI supports engineering decisions; it does not replace engineering accountability.
"""
)
|