phm-algo-ias 1.1.1__cp311-cp311-win_amd64.whl

algo/HI/__init__.py

@@ -0,0 +1,7 @@
+__version__ = "0.0.2"
+from . import HI
+from .HI import *
+
+__all__ = ["HI"]
+__all__ += HI.__all__
+# print("pythonds was called")

algo/RUL/__init__.py

@@ -0,0 +1,7 @@
+__version__ = "0.0.2"
+from . import RUL
+from .RUL import *
+
+__all__ = ["RUL"]
+__all__ += RUL.__all__
+# print("pythonds was called")

algo/__init__.py

@@ -0,0 +1,10 @@
+__version__ = '0.0.2'
+
+from . import HI
+from . import RUL
+# print('pythonds was called')
+
+try:
+    from ._version import version as __version__
+except Exception:  # 後備（理論上不太會用到）
+    __version__ = "0.0.0"

algo/_version.py

@@ -0,0 +1,34 @@
+# file generated by setuptools-scm
+# don't change, don't track in version control
+
+__all__ = [
+    "__version__",
+    "__version_tuple__",
+    "version",
+    "version_tuple",
+    "__commit_id__",
+    "commit_id",
+]
+
+TYPE_CHECKING = False
+if TYPE_CHECKING:
+    from typing import Tuple
+    from typing import Union
+
+    VERSION_TUPLE = Tuple[Union[int, str], ...]
+    COMMIT_ID = Union[str, None]
+else:
+    VERSION_TUPLE = object
+    COMMIT_ID = object
+
+version: str
+__version__: str
+__version_tuple__: VERSION_TUPLE
+version_tuple: VERSION_TUPLE
+commit_id: COMMIT_ID
+__commit_id__: COMMIT_ID
+
+__version__ = version = '1.1.1'
+__version_tuple__ = version_tuple = (1, 1, 1)
+
+__commit_id__ = commit_id = 'gbe7236c04'

phm_algo_ias-1.1.1.dist-info/METADATA

@@ -0,0 +1,19 @@
+Metadata-Version: 2.4
+Name: phm-algo-ias
+Version: 1.1.1
+Summary: Example algo package with Cython-compiled submodules
+Author: Your Name
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Cython
+Classifier: Programming Language :: C
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+Requires-Dist: passlib==1.7.4
+Requires-Dist: pandas==2.2.3
+Requires-Dist: numpy==2.0.1
+Requires-Dist: scipy==1.14.0
+Requires-Dist: scikit-learn==1.5.1
+Requires-Dist: matplotlib==3.9.2
+Requires-Dist: seaborn==0.13.2
+Requires-Dist: openpyxl==3.1.5
+Requires-Dist: statsmodels==0.14.4

phm_algo_ias-1.1.1.dist-info/RECORD

@@ -0,0 +1,12 @@
+algo/__init__.py,sha256=1rDS1k8TS4U4Q5ormDGtRkKkYIXyc8nGI1rJucpZxYo,238
+algo/_version.py,sha256=nrbsVx4IxWbEAZ8tUx6H-zY0_Oiv4NiXd48ZQuDPaFo,746
+algo/HI/HI.cp311-win_amd64.pyd,sha256=8BL8CJPAjBZFRmXoPFT0FAUT3cRs-zkB9kYU6GLe0AI,312320
+algo/HI/__init__.py,sha256=PbcO1CRc3T63WO8Uqtj3sw93wCalpOHQcbo6Va0KXro,133
+algo/RUL/RUL.cp311-win_amd64.pyd,sha256=2lLTLC0gFelWRw31y3NU38Pbb0bP7Ew5yMsFVdGGB0o,78848
+algo/RUL/__init__.py,sha256=g200G2bmP4feklvkSpcXfs0s3slO9dmRlWl5hXN6MIw,137
+test_code/HI_Verify.py,sha256=IWQunbEdFqkjYT41TZOMafEAu0eB_2Jw0siVgzSsimE,8856
+test_code/health_index_robot.py,sha256=pcI7K0yiyVXUL2-WY2vJY0KBNAn-EldlrNw3r_HBjmM,4602
+phm_algo_ias-1.1.1.dist-info/METADATA,sha256=7hEm24PWx2xb1Bj6NAiQua3SOAhIy-ebZGMVpv4cYmQ,629
+phm_algo_ias-1.1.1.dist-info/WHEEL,sha256=JLOMsP7F5qtkAkINx5UnzbFguf8CqZeraV8o04b0I8I,101
+phm_algo_ias-1.1.1.dist-info/top_level.txt,sha256=G1rxqg36Dwkx3w8o50708_IOrZelxPOJVB_KPOLlJM0,26
+phm_algo_ias-1.1.1.dist-info/RECORD,,

phm_algo_ias-1.1.1.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (80.9.0)
+Root-Is-Purelib: false
+Tag: cp311-cp311-win_amd64
+

phm_algo_ias-1.1.1.dist-info/top_level.txt

@@ -0,0 +1,3 @@
+algo
+test_code
+wheelhouse

test_code/HI_Verify.py

@@ -0,0 +1,220 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Mon Jul  3 14:19:40 2023
+
+@author: yunghua.chang
+"""
+
+import os
+import numpy as np
+import pandas as pd
+#import scipy.stats as st
+import datetime
+#from glob import glob
+#from matplotlib import pyplot as plt
+#from sklearn.preprocessing import StandardScaler
+#from sklearn.decomposition import PCA
+#from scipy.stats import f, skew
+#import matplotlib.ticker as ticker
+#import pickle
+import sys
+#from types import SimpleNamespace
+
+sys.path.append(os.getcwd())
+from algo.HI import HI
+from algo.RUL import RUL
+from datetime import timedelta
+
+import matplotlib.pyplot as plt
+import pickle   
+
+def plot(health, target):
+
+    health = health.copy()
+    health = health.sort_index()
+
+    # 用 datetime index 轉為字串來當作 x 軸標籤
+    health["TimeLabel"] = health.index.strftime("%Y-%m-%d %H:%M")
+
+    fig, ax = plt.subplots(figsize=(14, 5))
+    ax.plot(health["TimeLabel"], health["Score"], marker="o", color="#3A8FB7", label="Score")
+    ax.axhline(0.9 * 100, linestyle="--", color="#F26D5A", label="Threshold")
+
+    ax.set_title("Health Index - " + target, fontsize=20)
+    ax.set_ylabel("Score", fontsize=15)
+    ax.set_ylim(0, 100)
+
+    # 控制 x 軸顯示密度
+    step = max(len(health) // 12, 1)  # 至少每 12 個點顯示一次
+    ax.set_xticks(health["TimeLabel"][::step])
+    ax.set_xticklabels(health["TimeLabel"][::step], rotation=15, fontsize=10)
+
+    plt.yticks(fontsize=12)
+    plt.subplots_adjust(bottom=0.25)
+    plt.legend()
+    
+    plt.show()
+
+
+def drop_time_ranges(df, time_col="_time", ranges=None, fmt=None):
+    """
+    將 df 中 time_col 落在 ranges 內的列刪除（含起訖）。
+    ranges: [("YYYY/MM/DD HH:MM:SS", "YYYY/MM/DD HH:MM:SS"), ...]
+    fmt: 若你的時間字串有固定格式可填，否則留 None 讓 pandas 自動判斷
+    """
+    if ranges is None or len(ranges) == 0:
+        return df
+
+    out = df.copy()
+    # 轉成 datetime
+    out[time_col] = pd.to_datetime(out[time_col], format=fmt, errors="coerce")
+
+    for start, end in ranges:
+        start_dt = pd.to_datetime(start, format=fmt, errors="coerce")
+        end_dt   = pd.to_datetime(end,   format=fmt, errors="coerce")
+        mask = (out[time_col] >= start_dt) & (out[time_col] <= end_dt)
+        out = out.loc[~mask]
+
+    # 如果你後續還要把 Time 當 index 用：
+    # out = out.set_index(time_col).sort_index()
+
+    return out
+
+#----------------------------------- 改欄位名稱 -------------------------------------------------------------
+def rename_columns(df: pd.DataFrame) -> pd.DataFrame:
+    """
+    將輸入的 DataFrame 欄位依照對應表改名
+    Input: df (pandas.DataFrame)
+    Output: df (pandas.DataFrame, 欄位已改名)
+    """
+    # 先去掉欄位名稱前後空白
+    df.columns = df.columns.str.strip()
+
+    mapping_dict = {
+        "Time": "_time",
+        "RPM": "RPM",
+        "X-Overall": "X_Overall",
+        "Y-Overall": "Y_Overall",
+        "Z-Overall": "Z_Overall",
+        "X-Peak": "X_Peak",
+        "Y-Peak": "Y_Peak",
+        "Z-Peak": "Z_Peak",
+        "X-Peak to Peak": "X_Peak_to_Peak",
+        "Y-Peak to Peak": "Y_Peak_to_Peak",
+        "Z-Peak to Peak": "Z_Peak_to_Peak",
+        "X-Crest Factor": "X_Crest_Factor",
+        "Y-Crest Factor": "Y_Crest_Factor",
+        "Z-Crest Factor": "Z_Crest_Factor",
+
+        "X-Power in Band 01": "X_Power_in_Band_01",
+        "X-Power in Band 02": "X_Power_in_Band_02",
+        "X-Power in Band 03": "X_Power_in_Band_03",
+        "X-Power in Band 04": "X_Power_in_Band_04",
+        "X-Power in Band 05": "X_Power_in_Band_05",
+        "X-Power in Band 06": "X_Power_in_Band_06",
+        "X-Power in Band 07": "X_Power_in_Band_07",
+        "X-Power in Band 08": "X_Power_in_Band_08",
+        "X-Power in Band 09": "X_Power_in_Band_09",
+        "X-Power in Band 10": "X_Power_in_Band_10",
+
+        "Y-Power in Band 01": "Y_Power_in_Band_01",
+        "Y-Power in Band 02": "Y_Power_in_Band_02",
+        "Y-Power in Band 03": "Y_Power_in_Band_03",
+        "Y-Power in Band 04": "Y_Power_in_Band_04",
+        "Y-Power in Band 05": "Y_Power_in_Band_05",
+        "Y-Power in Band 06": "Y_Power_in_Band_06",
+        "Y-Power in Band 07": "Y_Power_in_Band_07",
+        "Y-Power in Band 08": "Y_Power_in_Band_08",
+        "Y-Power in Band 09": "Y_Power_in_Band_09",
+        "Y-Power in Band 10": "Y_Power_in_Band_10",
+
+        "Z-Power in Band 01": "Z_Power_in_Band_01",
+        "Z-Power in Band 02": "Z_Power_in_Band_02",
+        "Z-Power in Band 03": "Z_Power_in_Band_03",
+        "Z-Power in Band 04": "Z_Power_in_Band_04",
+        "Z-Power in Band 05": "Z_Power_in_Band_05",
+        "Z-Power in Band 06": "Z_Power_in_Band_06",
+        "Z-Power in Band 07": "Z_Power_in_Band_07",
+        "Z-Power in Band 08": "Z_Power_in_Band_08",
+        "Z-Power in Band 09": "Z_Power_in_Band_09",
+        "Z-Power in Band 10": "Z_Power_in_Band_10",
+
+        "Bearing": "Bearing",
+        "Looseness": "Looseness",
+        "Misalignment": "Misalignment",
+        "Imbalance": "Imbalance",
+        "Gear_mesh": "Gear_mesh",
+        "Van_Pass": "Van_Pass",
+
+        "Bearing %": "Bearing_percentage",
+        "Looseness %": "Looseness_percentage",
+        "Misalignment %": "Misalignment_percentage",
+        "Imbalance %": "Imbalance_percentage",
+        "Gear_mesh %": "Gear_mesh_percentage",
+        "Van_Pass %": "Van_Pass_percentage",
+
+        "Temperature": "Temperature",
+        "X-Overall QC": "X_Overall_QC",
+        "Y-Overall QC": "Y_Overall_QC",
+        "Z-Overall QC": "Z_Overall_QC",
+    }
+
+    df = df.rename(columns=mapping_dict, errors="ignore")
+    return df
+
+if __name__ == "__main__":
+    #------------------------------- << Training Model >> ---------------------------------------------------------
+    #time_scale = "h" # 輸入時間維度 : "每小時"("h"), "每分鐘"("min")
+    file_path = os.path.join(os.getcwd(), "data", "410_good_1_all_feature.csv")
+    df = pd.read_csv(file_path) 
+
+    #df = df[["Time"] + df.select_dtypes(include="number").columns.tolist()]
+    df = rename_columns(df) # 更改欄位名稱 (.CSV檔欄位名稱->系統欄位名稱)
+
+    algo_config = "1, 0, 1, 1, 2, 1, 1" # [時長維度, testing資料離處理, 低解析度特徵刪除, 特徵選擇, 資料正規化, 模型, rul_deadline]
+
+    result = HI.training_model(df, algo_config) # Training Model : "training_model_robust_index"(訓練模型指標分數), "training_model_robust_status"(verify:綠燈，warning:紅燈)
+    print(result)
+
+    # result_clean = HI.outlier_clean(df, result) # 測試資料清洗後的model
+    # print(result_clean)
+
+    #Health_train = plot(result['HI_Score'], target="train") # 繪出HI圖
+
+    #展覽pickle檔案輸出
+    # pick_path = os.path.join(os.getcwd(), "uploaded_files", "model", "model.pkl")
+    # to_save = dict(result)
+    # #to_save["for_demo"] = "for_demo" # 展覽pickle輸出，給系統辨識
+    # with open(pick_path, "wb") as f:
+    #     pickle.dump(to_save, f)
+
+    #----------------------------------------<<測試一好一壞算法>>---------------------------------------------------------
+    # file_path_1 = os.path.join(os.getcwd(), "data", "410_good_1_all_feature.csv")
+    # df_1 = pd.read_csv(file_path_1) 
+    # df_1 = rename_columns(df_1)
+    # file_path_2 = os.path.join(os.getcwd(), "data", "Robust_Bad.csv")
+    # df_2 = pd.read_csv(file_path_2) 
+    # df_2 = rename_columns(df_2)
+    # good_bad_result = HI.training_model_GB_compare(df_1, df_2,  time_scale)
+
+    #------------------------------- << Inference_HI & 嫌疑度變量 >> ---------------------------------------------------------
+    file_path_test = os.path.join(os.getcwd(), "data", "410_good_1_all_feature.csv")
+    test_df = pd.read_csv(file_path_test) 
+    test_df = rename_columns(test_df) # 更改欄位名稱 (.CSV檔欄位名稱->系統欄位名稱)
+    test_result = HI.testing_model(test_df,result) # Inference : 輸出  "HI_Score"(HI分數), "suspicious_variable"(變量嫌疑度)
+    print(test_result)
+
+    #test_cleaned = HI.testing_model(test_df,result_clean) # 測試資料清洗後的model
+    #Health_test = plot(test_result['HI_Score'], target="test")
+
+    #------------------------------------- << RUL >> ------------------------------------------------------------------------
+    RUL_result = RUL.rul_calculation_h(test_result["HI_Score"], result) # Inference : 輸出 RUL
+    print(RUL_result)
+
+    #------------------------------- << HI_Score 異常時間區段 >> --------------------------------------------------------------
+    fail_time = HI.detect_bad_HI_zone(test_result["HI_Score"],threshold=90)
+    print(fail_time)
+
+
+
+

test_code/health_index_robot.py

@@ -0,0 +1,178 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Mon Jul  3 14:19:40 2023
+
+@author: yunghua.chang
+"""
+
+import os
+import numpy as np
+import pandas as pd
+import scipy.stats as st
+import datetime
+from glob import glob
+from matplotlib import pyplot as plt
+from sklearn.preprocessing import StandardScaler
+from sklearn.decomposition import PCA
+from scipy.stats import f, skew
+import matplotlib.ticker as ticker
+
+#%% Health Index計算
+def health_modeling(df, score, alpha=0.05):
+    # 標準化
+    scaler = StandardScaler()
+    scaler.fit(df)
+    df_scaled = scaler.transform(df)
+
+    # 進行主成份分析降維
+    pca = PCA(n_components=0.9)
+    pca.fit(df_scaled)
+    df_pca = pca.transform(df_scaled)
+
+    # 計算模型的 mu & sigma
+    model_mu = df_pca.mean(axis=0)
+    model_cov = np.cov(df_pca.T)
+
+    # 計算模型的UCL
+    m = df_pca.shape[0]
+    p = df_pca.shape[1]
+    fscore = f.ppf(1 - alpha, p, m - p)
+    ucl = (p * (m - 1)) / (m - p) * fscore
+
+    ## 計算分數
+    df_health = Health_Index(df_pca, model_mu, model_cov, ucl, df.index, score)
+
+    return df_health, model_mu, model_cov, ucl, scaler, pca
+
+
+def health_testing(df, model_mu, model_cov, ucl, scaler, pca):
+
+    # 以模型平均&標準差進行標準
+    df_scaled = scaler.transform(df)
+    # 進行主成份分析降維
+    df_pca = pca.transform(df_scaled)
+    ## 計算分數
+    df_health = Health_Index(df_pca, model_mu, model_cov, ucl, df.index, score)
+
+    return df_health
+
+
+def score_ewma(df_health, score):
+    ## EWMA部分
+    grade_rolling = df_health["Score"].ewm(span=3, adjust=False).mean()
+    index = grade_rolling[df_health["Score"] < score * 100].index
+    df_health.loc[index, "Score"] = grade_rolling[
+        df_health["Score"] < score * 100
+    ]
+    return df_health
+
+
+def CalTSquare(data_pca, i, mu, sigma):
+    x = data_pca[i]
+    bias = x - mu
+    tsquare = bias.dot(np.linalg.pinv(sigma)).dot(np.transpose(bias))
+    return tsquare
+
+
+def CalScore(tsquare, ucl, score):
+
+    score_ = np.exp(np.log(score) * ((tsquare) / ucl)) * 100
+
+    return score_
+
+
+def Health_Index(data_pca, model_mu, model_cov, ucl, index, score):
+
+    df = pd.DataFrame(index=index)
+    df["Tsquare"] = list(
+        map(
+            lambda x: CalTSquare(data_pca, x, model_mu, model_cov),
+            np.arange(len(data_pca)),
+        )
+    )
+    df["Score"] = df["Tsquare"].apply(lambda x: CalScore(x, ucl, score))
+
+    return df
+
+
+def plot(health, target):
+
+    tick_spacing = 15
+    fig, ax = plt.subplots(figsize=(20, 5))
+    ax.plot(health.index, health["Score"], marker="o")
+    ax.axhline(score * 100, color="r")
+    ax.xaxis.set_major_locator(ticker.MultipleLocator(tick_spacing))
+    plt.title(" Health Index ," + target, fontsize=25)
+    plt.yticks(fontsize=20)
+    plt.xticks(rotation=15, fontsize=20)
+    plt.ylim(0, 100)
+    plt.show()
+
+
+
+#%% 參數設定
+score = 0.9
+modeling_start_time = "2024-09-30 16:00:00"
+modeling_end_time = "2024-10-21 23:59:59"
+
+file_path = r"C:\Users\chiyu.hong\Desktop\T7_TLRB0100_00_00001_____9_2_edc_raw_data.csv"
+
+df = pd.read_csv(file_path) 
+
+df["txn_dttm"] = pd.to_datetime(df["txn_dttm"], errors="coerce")
+df = df.set_index("txn_dttm")
+
+indicator_name = [
+    "X-OA_MAX~_X",	
+    "Y-OA_MAX~_X",	
+    "Z-B01_MAX~_X",	
+    "Z-OA_MAX~_X",	
+    "Y-B03_MAX~_X",	
+    "X-B03_MAX~_X",	
+    "TSquare~_X",	
+    "Y-B01_MAX~_X",	
+    "Z-B02_MAX~_X",	
+    "X-B01_MAX~_X",	
+    "Z-B03_MAX~_X",	
+    "Y-B02_MAX~_X",	
+    "X-B02_MAX~_X"
+]
+
+#%%
+std_name = list(map(lambda x: x + "_std", indicator_name))
+mean_name = list(map(lambda x: x + "_mean", indicator_name))
+
+#%%
+df_copy = df.copy()
+df = df_copy[indicator_name].resample("H").max()
+df = df.dropna(axis="index", how="any")
+df[mean_name] = df_copy[indicator_name].resample("H").mean()
+df[std_name] = df_copy[indicator_name].resample("H").std()
+
+df = df.dropna(axis="index", how="any")
+
+    
+
+if __name__ == "__main__":
+    target = "Train"
+
+    (df_health_train, mu, cov, ucl, scaler, pca) = health_modeling(
+        df.loc[modeling_start_time:modeling_end_time], score
+    )
+
+    df_health_train = df_health_train.rolling("3H").mean()
+
+    # Health_train = plot(df_health_train, target)
+
+#%%
+## Test
+if __name__ == "__main__":
+    target = "Test"
+    df_health_test = health_testing(
+        df.loc[modeling_end_time:], mu, cov, ucl, scaler, pca
+    )
+
+    df_health_test = df_health_test.rolling("3H").mean()
+
+    Health_test = plot(df_health_test, target)
+