Update app.py

app.py

@@ -18,7 +18,7 @@ def gen_pulsive_noise(count, min_val, max_val):
         pulse = random.randint(min_val, max_val)
     
     return pulse
-def simulation(input_noise, output_noise, min, max, std, mean, min_pul, max_pul, kp, ki, kd, show_original):
+def simulation(input_noise, output_noise, min, max, std, mean, min_pul, max_pul, kd, show_original, derivative_control, moving_average, window_size):
     vrs = {}
     fmu = 'Pharmacokinetics_4_comportmental_model_PI_ref_FMU_base4_OAAS_lnx.fmu'
     model_description = read_model_description(fmu)
@@ -60,6 +60,11 @@ def simulation(input_noise, output_noise, min, max, std, mean, min_pul, max_pul,
     i = 0
     target = 30
     last_error = 0
+    last_bis = 0
+    bis_history = []
+    kp = 4
+    ki = 0.01
+    kd = kd if derivative_control else 0
     # simulation loop
     while time < stop_time:
 
@@ -75,6 +80,9 @@ def simulation(input_noise, output_noise, min, max, std, mean, min_pul, max_pul,
         bis = fmu.getReal([int(vr_output)])[0] if time > step_size else 95.6
         if input_noise:
             bis += add_noise(min, max, std, mean)
+        if moving_average:
+            bis_history.append(bis)
+            bis = bis if time <= window_size else np.mean(bis_history[-window_size:])
         p = bis - target
         i = i + p
         d = p - last_error
@@ -100,9 +108,7 @@ def simulation(input_noise, output_noise, min, max, std, mean, min_pul, max_pul,
     fmu.freeInstance()
     shutil.rmtree(unzipdir, ignore_errors=True)
     result = np.array(rows, dtype=np.dtype([('time', np.float64), ('BIS', np.float64), ('Infusion', np.float64)]))
-    result_baseline = np.load("result.npy")
     df = pd.DataFrame(result)  
-    df_original = pd.DataFrame(result_baseline)
     trace1 = go.Scatter(x=df.index, y=df['BIS'], mode='lines', name='BIS')
     fig1 = go.Figure(data=trace1)
     fig1.update_layout(height=400, width=1200, title_text="BIS evolution")
@@ -111,8 +117,12 @@ def simulation(input_noise, output_noise, min, max, std, mean, min_pul, max_pul,
     trace2 = go.Scatter(x=df.index, y=df['Infusion'], mode='lines', name='Infusion')
     fig2 = go.Figure(data=trace2)
     fig2.update_layout(height=400, width=1200, title_text="Infusion evolution")
-    
     if show_original:
+        if input_noise:
+            result_baseline = np.load("result_gaussian.npy")
+        elif output_noise:
+            result_baseline = np.load("result_pulse.npy")
+        df_original = pd.DataFrame(result_baseline)
         fig1.add_trace(go.Scatter(x=df_original.index, y=df_original['BIS'], mode='lines', name='BIS original', line=dict(color="red"), opacity=0.5))
         fig2.add_trace(go.Scatter(x=df_original.index, y=df_original['Infusion'], mode='lines', name='Infusion original',  line=dict(color="red"), opacity=0.5))
 
@@ -123,7 +133,7 @@ with gr.Blocks() as demo:
         with gr.Column(scale=1):
             gr.Markdown("**BIS noise**")
             input_noise = gr.inputs.Checkbox(label="Add Gaussian noise")
-            with gr.Accordion("noise range"):
+            with gr.Accordion("noise range", default_open=False):
                     min_gaussian = gr.inputs.Slider(minimum=0, maximum=20, step=1, default=5, label="noise min")
                     max_gaussian = gr.inputs.Slider(minimum=0, maximum=20, step=1, default=10, label="noise max")
                     std_gaussian = gr.inputs.Slider(minimum=0, maximum=10, step=1, default=2, label="noise standard deviation")
@@ -131,18 +141,22 @@ with gr.Blocks() as demo:
             gr.Markdown("**Infusion noise**")
             with gr.Blocks():
                 output_noise = gr.inputs.Checkbox(label="Add Pulsive noise")
-                with gr.Accordion("noise range"):
+                with gr.Accordion("noise range", default_open=False):
                     min_pul = gr.inputs.Slider(minimum=0, maximum=50, step=1, default=50, label="noise min")
                     max_pul = gr.inputs.Slider(minimum=0, maximum=150, step=1, default=150, label="noise max")
-            gr.Markdown("**PID controller paramters**")
-            with gr.Blocks():
-                kp_slider = gr.inputs.Slider(minimum=0, maximum=20, default=4, label="kp")
-                ki_slider = gr.inputs.Slider(minimum=0, maximum=1, default=0.01, label="ki")
-                kd_slider = gr.inputs.Slider(minimum=0, maximum=200, default=120, label="kd")
-            show_original = gr.inputs.Checkbox(label="Show original")
+            gr.Markdown("**Advanced control**")
+            with gr.Tab("Smoothing algorithm"):
+                with gr.Row():
+                    moving_average = gr.inputs.Checkbox(label="Moving average filter")
+                    window_size = gr.inputs.Slider(minimum=0, maximum=200, default=0, label="Window size")
+            with gr.Tab("Derivative control"):
+                with gr.Row():
+                    derivative_control = gr.inputs.Checkbox(label="Adding derivative term to PI controller")
+                    kd_slider = gr.inputs.Slider(minimum=0, maximum=200, default=0, label="kd")
             button = gr.Button("Simulate")
+            show_original = gr.inputs.Checkbox(label="Show original")
         with gr.Column(scale=5):
             plot1 = gr.Plot(label="BIS evolution")
             plot2 = gr.Plot(label="Infusion evolution")
-    button.click(simulation, inputs=[input_noise, output_noise, min_gaussian, max_gaussian, std_gaussian, mean_gaussian, min_pul, max_pul, kp_slider, ki_slider, kd_slider, show_original], outputs=[plot1, plot2])
+    button.click(simulation, inputs=[input_noise, output_noise, min_gaussian, max_gaussian, std_gaussian, mean_gaussian, min_pul, max_pul, kd_slider, show_original, derivative_control, moving_average, window_size], outputs=[plot1, plot2])
 demo.launch()