Update app.py

app.py

@@ -1,74 +1,88 @@
-import streamlit as st
-import joblib
-import pandas as pd
 import numpy as np
 import matplotlib.pyplot as plt
-from models.optimizer import optimize_design
-# Load pre-trained model
-model = joblib.load('models/defect_model.pkl')
-# Page Title
-19/112
-st.title("AI-Driven Press Tool Defect Predictor & Optimizer")
-# Sidebar: User Input Parameters
-st.sidebar.header("Input Parameters")
-tool_width = st.sidebar.slider("Tool Width (mm)", min_value=30, max_value=100,
-value=50, step=5)
-tool_height = st.sidebar.slider("Tool Height (mm)", min_value=10, max_value=50,
-value=20, step=2)
-material_strength = st.sidebar.slider("Material Strength (MPa)", min_value=200,
-max_value=500, value=300, step=20)
-press_force = st.sidebar.slider("Press Force (kN)", min_value=50, max_value=300,
-value=100, step=10)
-# Input Data Frame
-input_data = pd.DataFrame({
-'tool_width': [tool_width],
-'tool_height': [tool_height],
-'material_strength': [material_strength],
-'press_force': [press_force]
-})
-# Predict Defect
-defect = model.predict(input_data)[0]
-probabilities = model.predict_proba(input_data)[0]
-# Optimize Design
-optimized_params = optimize_design(tool_width, tool_height, material_strength,
-press_force)
-# Main Section: Results
-st.subheader("Defect Prediction and Optimization Results")
-# Prediction Result
-st.markdown(f"### Predicted Defect: **{defect}**")
-# Visualize Prediction Probabilities
-st.markdown("### Prediction Probabilities")
-fig, ax = plt.subplots()
-labels = model.classes_
-ax.bar(labels, probabilities, color='skyblue')
-ax.set_ylabel("Probability")
-ax.set_title("Defect Prediction Probabilities")
-20/112
-Enhancements Added
-1. Bar Chart for Defect Probabilities:
-Displays the condence level for each defect prediction.
-2. Real-Time Parameter Adjustment:
-Adjust parameters like tool width, height, material strength, and press force using
-sliders.
-Instantly see how predictions and optimizations are aected.
-3. Line Chart for Parameter Impact:
-Visualizes the relationship between tool width and defect probabilities.
-Similar visualizations can be added for other parameters.
-st.pyplot(fig)
-# Optimized Parameters
-st.markdown("### Optimized Design Parameters")
-st.json(optimized_params)
-# Dynamic Visualization: Adjusting Parameters
-st.markdown("### Parameter Impact on Predictions")
-impact_chart_data = []
-for w in range(30, 101, 10):
-temp_data = pd.DataFrame({
-'tool_width': [w],
-'tool_height': [tool_height],
-'material_strength': [material_strength],
-'press_force': [press_force]
-})
-prob = model.predict_proba(temp_data)[0]
-impact_chart_data.append((w, prob))
-impact_chart_df = pd.DataFrame(impact_chart_data, columns=["Tool Width", *labels])
-st.line_chart(impact_chart_df.set_index("Tool Width"))
+from ansys.mapdl.core import launch_mapdl
+# Launch ANSYS MAPDL instance
+mapdl = launch_mapdl()
+# Clear any previous session
+mapdl.clear()
+mapdl.prep7() # Enter preprocessor
+# Function to define and simulate the boring machine attachment
+def simulate_attachment(thickness):
+# Define material properties
+mapdl.mp("EX", 1, 2e11) # Elastic modulus (Pa)
+mapdl.mp("PRXY", 1, 0.3) # Poisson's ratio
+# Create geometry
+mapdl.block(0, 200, 0, 100, 0, thickness) # Rectangular block
+mapdl.cylind(0, 10, 50, 50, 0, thickness) # Cylindrical hole
+mapdl.vsubtract("ALL") # Subtract cylinder from block
+# Define element type and mesh
+mapdl.et(1, "SOLID185") # 3D structural element
+mapdl.esize(5) # Element size
+mapdl.vmesh("ALL") # Mesh all volumes
+# Apply boundary conditions
+mapdl.nsel("S", "LOC", "X", 0) # Select nodes on one end
+38/112
+mapdl.d("ALL", "ALL") # Fix all degrees of freedom
+mapdl.nsel("S", "LOC", "X", 200) # Select nodes on the other end
+mapdl.f("ALL", "FY", -5000) # Apply force
+# Solve
+mapdl.run("/SOLU") # Enter solution phase
+mapdl.antype("STATIC") # Static analysis
+mapdl.solve()
+mapdl.finish() # Finish solution
+# Post-process
+mapdl.post1()
+max_stress = mapdl.get_value("NODE", 0, "S", "EQV") # Equivalent stress
+max_def = mapdl.get_value("NODE", 0, "U", "SUM") # Total deformation
+# Save plots
+mapdl.post_processing.plot_nodal_solution(
+"S", "EQV", title=f"Stress Distribution (Thickness={thickness} mm)",
+savefig=f"stress_{thickness}.png"
+)
+mapdl.post_processing.plot_nodal_solution(
+"U", "SUM", title=f"Deformation (Thickness={thickness} mm)",
+savefig=f"deformation_{thickness}.png"
+)
+return max_stress, max_def
+# Parameter sweep for different thicknesses
+thickness_values = [20, 25, 30]
+results = []
+for thickness in thickness_values:
+max_stress, max_def = simulate_attachment(thickness)
+results.append({"Thickness (mm)": thickness, "Max Stress (Pa)": max_stress, "Max
+Deformation (mm)": max_def})
+# Visualize results
+import pandas as pd
+results_df = pd.DataFrame(results)
+print(results_df)
+39/112
+Key Features
+1. Geometry Creation:
+Python creates the geometry of the attachment, including a rectangular block with a
+cylindrical hole.
+2. Material Properties:
+Elastic modulus and Poisson's ratio are assigned programmatically.
+3. Meshing:
+Element type SOLID185 and ne mesh control are applied.
+4. Boundary Conditions and Loading:
+Fixes one side and applies a distributed load on the opposite side.
+5. Result Extraction:
+Maximum stress and deformation are retrieved for each thickness.
+Results are saved as plots for stress distribution and deformation.
+6. Parameter Sweep:
+Thickness is varied (20 mm, 25 mm, 30 mm), and simulations are run for each case.
+# Plot results
+plt.figure(figsize=(10, 5))
+plt.plot(results_df["Thickness (mm)"], results_df["Max Stress (Pa)"], marker="o",
+label="Max Stress")
+plt.plot(results_df["Thickness (mm)"], results_df["Max Deformation (mm)"],
+marker="o", label="Max Deformation")
+plt.xlabel("Thickness (mm)")
+plt.ylabel("Value")
+plt.title("Effect of Thickness on Stress and Deformation")
+plt.legend()
+plt.grid()
+plt.savefig("results_plot.png")
+plt.show()
+# Exit MAPDL
+mapdl.exit()