wip bess swap time prediction

Utilities/BESS.py

@@ -1,3 +1,6 @@
+import pandas as pd
+
+
 def initialise_SoC(bess):
     """Initialise the state of charge (SoC) for the BESS."""
     for i in range(0, len(bess["units"])):  # initially fully charged
@@ -7,13 +10,14 @@ def initialise_SoC(bess):
 
 def initial_site_assignment(c, bess):
     """Initialise the site assignment for each BESS."""
-    k = 0
-    while k < len(c["site_info"]["sites"]):
-        bess["units"][k]["site"] = c["site_info"]["sites"][k]["name"]
-        k += 1
 
+    for k in range(0, len(bess["units"])):
+        # assign each BESS unit to a site
         if k < len(c["site_info"]["sites"]):
+            bess["units"][k]["site"] = c["site_info"]["sites"][k]["name"]
+        else:
             bess["units"][k]["site"] = "Unassigned"
+
     return bess
 
 
@@ -23,10 +27,14 @@ def discharge_bess(bess, site_name, dt, discharge_power):
 
     """Discharge the BESS for a specific site."""
     for index, unit in enumerate(bess["units"]):
+        if unit["site"] == "Unassigned":
+            continue
+
         if unit["site"] == site_name:
             new_soc = unit["SoC"] - (dt * discharge_energy) / unit["capacity_kWh"]
             new_soc = 0 if new_soc < 0 else new_soc
         else:
+            # maintain SoC if not assigned to the site
             new_soc = unit["SoC"]
 
         # update SoC
@@ -37,8 +45,49 @@ def discharge_bess(bess, site_name, dt, discharge_power):
 def predict_swap_time(bess_soc_for_cycle):
     """Predict the swap time for each BESS unit based on its SoC history."""
     swap_times = {}
+    min2sec = 60
+    threshold = 2 * min2sec  # 2 minutes in seconds
+
     for unit_name, df in bess_soc_for_cycle.items():
-        # Find the timestamp when SoC reaches 0
+        # need to be at least 1 min of operation to start estimation
-        swap_time = df[df["SoC"] == 0]["Timestamp"].min()
+        if len(df) < threshold:
+            swap_times[unit_name] = None
+            continue
+
+        # linear extrapolation to estimate swap time
+        # calculate the slope of the SoC over time
+        m = (df["SoC"].iloc[-1] - df["SoC"].iloc[0]) / (
+            df["Timestamp"].iloc[-1] - df["Timestamp"].iloc[0]
+        )
+
+        if m == 0:
+            swap_times[unit_name] = None
+            continue
+
+        # solve for the time when SoC reaches 0
+        swap_time = (0 - df["SoC"].iloc[0]) / m + df["Timestamp"].iloc[0]
+        # assign to swap_times
         swap_times[unit_name] = swap_time
+
     return swap_times
+
+
+def update_cycle_SoC(bess_data, bess_soc_for_cycle, timestamps):
+    init_df = pd.DataFrame(columns=["Timestamp", "SoC"])
+    # assign SoC for cycle
+    for unit in bess_data["units"]:
+        unit_name = unit["name"]
+        # reset df if SoC is 0. Start a new cycle
+        if unit["SoC"] == 0:
+            bess_soc_for_cycle[unit_name] = init_df
+
+        bess_soc_for_cycle[unit_name] = pd.concat(
+            [
+                bess_soc_for_cycle[unit_name],
+                pd.DataFrame(
+                    [[timestamps[i], unit["SoC"]]],
+                    columns=["Timestamp", "SoC"],
+                ),
+            ],
+            axis=0,
+        )

main.py

@@ -7,6 +7,7 @@ from Utilities.BESS import (
     initial_site_assignment,
     discharge_bess,
     predict_swap_time,
+    update_cycle_SoC,
 )
 import matplotlib.pyplot as pl
 import pandas as pd
@@ -95,22 +96,9 @@ with ThreadPoolExecutor() as executor:
             axis=0,
         )
 
-        # assign SoC for cycle
+        # update cycle SoC
-        for unit in bess_data["units"]:
+        # this is for predicting swap times
-            unit_name = unit["name"]
+        bess_soc_for_cycle = update_cycle_SoC(bess_data, bess_soc_for_cycle, timestamps)
-            # reset df if SoC is 0. Start a new cycle
-            if unit["SoC"] == 0:
-                bess_soc_for_cycle[unit_name] = init_df
-            bess_soc_for_cycle[unit_name] = pd.concat(
-                [
-                    bess_soc_for_cycle[unit_name],
-                    pd.DataFrame(
-                        [[timestamps[i], unit["SoC"]]],
-                        columns=["Timestamp", "SoC"],
-                    ),
-                ],
-                axis=0,
-            )
 
         # predict swap times
         swap_times = predict_swap_time(bess_soc_for_cycle)