Module ctsimu.evaluation.testTwin
Test of digital model
To run this test evaluation, you can simply pass the name of the test and the metadata file which describes the data files that you want to evaluate to the Toolbox:
from ctsimu.toolbox import Toolbox
Toolbox("twin", "TwinTest_metadata.json")
The metafile for the CTSimU Twin Test has the following structure:
{
"file": {
"file_type": "CTSimU Twin Test", # do not change
"file_format_version": {"major": 0, "minor": 4} # do not change
},
"measurement": {
"description": "DigitalTwinTest on sample data using Lochplatte.", # in report
"contact": "Jane Doe", # in report
"sample_material": "Al", # not used, for information
"material_alpha": 0.0000234, # required, thermal expansion coefficient
"measurands": ["dia01", "dia02", "dia03", "dia04", "dia05"],
# required, column names
"output_path": "./results", # required
"csv_sep": ";", # required, column separator
"csv_decimal": "," # required, decimal mark
},
"values-real": {
"description": "CT XYZ", # in report
"nr_of_runs":"", # not used
"files": "", # not used, instead take all CSVs at 'csv_path'
"csv_path": "Data/Measurements/", # required
"csv_sep": ";", # required, column separator
"csv_decimal": ",", # required, decimal mark
"header_row": 103, # required
"name_column": "Name", # required
"value_column": "Ist [mm/°]", # required
"k_value": 2, # required
"temperature": 21.3, # required
"scaling_factor": 1.0 # required
},
"values-sim":{
"description": "simulation XYZ", # in report
"nr_of_runs":"", # not used
"files": "", # not used, instead take all CSVs at 'csv_path'
"csv_path": "Data/Simulations/", # required
"csv_sep": ";", # required, column separator
"csv_decimal": ",", # required, decimal mark
"header_row": 103, # required
"name_column": "Name", # required
"value_column": "Ist [mm/°]", # required
"temperature": 20, # required
"scaling_factor": 1 # required
},
"reference-real":{
"description": "Calibration XYZ", # in report
"temperature_calib": 20, # required
"csv_path": "Daten/Calibration.csv", # required
"csv_sep": ";", # required, column separator
"csv_decimal": ",", # required, decimal mark
"name_column": "", # required
"value_column": "Kalibrierwert (mm)", # required
"uncertainty_column": "Kalibrierunsicherheit" # required
},
"reference-sim":{
"description": "Calibration XYZ", # in report
"temperature_calib": 20, # required
"csv_path": "Daten/Calibration.csv", # required
"csv_sep": ";", # required, column separator
"csv_decimal": ",", # required, decimal mark
"name_column": "", # required
"value_column": "Nominal STL", # required
"uncertainty_column": "" # required
}
}
The "measurands" are defined as a list of identifiers. An entry can end with a placeholder '*', which activates all identifiers that begin with the character string before the placeholder. If there is an entry with a placeholder in the list, entries without a placeholder are ignored.
Classes
class testTwin (filename: str = None)-
CTSimU2 test of digital twin.
Expand source code
class testTwin(): """ CTSimU2 test of digital twin. """ def __init__(self, filename:str=None): # empty declarations self.Criterion_real = [] self.Criterion_sim = [] self.combined_df = [] self.df = pd.DataFrame() self.img_buf = io.BytesIO() # metadatafile --> Json File #self.metadatafile = metadatafile #self.metadata = read_json_file(self.metadatafile) self.metadata = {} self.current_metadata_path = None self.current_metadata_file = None self.current_metadata_basename = None self.current_metadata_directory = None self.metadata_is_set = False if filename is not None: self.read_metadata(filename) def reset_metadata(self): """Reset scenario's metadata information.""" self.current_metadata_path = None self.current_metadata_file = None self.current_metadata_basename = None self.current_metadata_directory = None self.metadata_is_set = False # Create new, empty metadata: self.metadata = {} def read_metadata(self, filename:str=None): """Import metadata from a CTSimU Twin Test file or a given metadata dictionary. Parameters ---------- filename : str Path to a CTSimU Twin Test file. Default value: `None` """ if filename is not None: json_dict = read_json_file(filename=filename) self.current_metadata_path = filename self.current_metadata_directory = os.path.dirname(filename) self.current_metadata_file = os.path.basename(filename) self.current_metadata_basename, extension = os.path.splitext(self.current_metadata_file) self.name = self.current_metadata_basename # If a file is read, we want to make sure that it is a valid # and supported metadata file: if isinstance(json_dict, dict): file_type = get_value(json_dict, ["file", "file_type"]) if file_type != "CTSimU Twin Test": raise Exception(f"Invalid metadata structure: the string 'CTSimU Twin Test' was not found in 'file.file_type' in the metadata file {filename}.") fileformatversion = get_value(json_dict, ["file", "file_format_version"]) if not is_version_supported(ctsimu_twin_test_supported_version, fileformatversion): raise Exception(f"Unsupported or invalid metadata version. Currently supported: up to {ctsimu_twin_test_supported_version['major']}.{ctsimu_twin_test_supported_version['minor']}.") else: raise Exception(f"Error when reading the metadata file: {filename}. read_json_file did not return a Python dictionary.") self.metadata = json_dict self.metadata_is_set = True #general Information for the test self.measurands = self.metadata['measurement']['measurands'] self.output_path = self.metadata['measurement']['output_path'] self.sep = self.metadata['measurement']['csv_sep'] self.decimal = self.metadata['measurement']['csv_decimal'] self.alpha = self.metadata['measurement']['material_alpha'] def prepare_data(self): for table_id in table_ids[::-1]: print(table_id) folder_path = self.metadata[table_id]['csv_path'] sep = self.metadata[table_id]['csv_sep'] decimal = self.metadata[table_id]['csv_decimal'] name_col = self.metadata[table_id]['name_column'] value_col = self.metadata[table_id]['value_column'] if table_id == table_ids[2]: # 'reference-real' self.RealRefT = self.metadata[table_id]['temperature_calib'] data = pd.read_csv(folder_path, sep=sep, decimal=decimal, header=0, index_col=0) self.RealValuesRef = data[value_col] uncertaintyCol = self.metadata[table_id]['uncertainty_column'] self.RealRefUncertainty = data[uncertaintyCol] # Expand list of measurands if len(self.measurands) == 0: # take all calibrated quantities of the test piece self.measurands = data.index elif len([m for m in self.measurands if '*' in m]) > 0: # wildcard patterns only mds = [] for m in data.index: for pattern in [m for m in self.measurands if '*' in m]: if m.startswith(pattern.replace('*', "")): mds.append(m) self.measurands = mds print(self.measurands) elif table_id == table_ids[3]: # 'reference-sim' self.SimRefT = self.metadata[table_id]['temperature_calib'] data = pd.read_csv(folder_path, sep=sep, decimal=decimal, header=0, index_col=0) self.SimValuesRef = data[value_col] # uncertaintyCol = self.metadata[table_id]['uncertainty_column'] # self.CalUncertainty = self.data[uncertaintyCol] elif table_id == table_ids[0]: self.RealValues = self.read_and_filter_csv_files(folder_path, table_id) elif table_id == table_ids[1]: self.SimValues = self.read_and_filter_csv_files(folder_path, table_id) def read_and_filter_csv_files(self, folder_paths, ct_type): sep = self.metadata[ct_type]['csv_sep'] decimal = self.metadata[ct_type]['csv_decimal'] header_row = self.metadata[ct_type]['header_row'] name_col = self.metadata[ct_type]['name_column'] value_col = self.metadata[ct_type]['value_column'] temperature = self.metadata[ct_type]['temperature'] scaling_factor = self.metadata[ct_type]["scaling_factor"] # Choose calibration if ct_type == "values-sim": ref_values = self.SimValuesRef ref_temperature = self.SimRefT else: ref_values = self.RealValuesRef ref_temperature = self.RealRefT combined_df_test = pd.DataFrame() file_names = [] for folder_path in [folder_paths]: for filename in os.listdir(folder_path): if filename.endswith('.csv'): file_path = os.path.join(folder_path, filename) df = pd.read_csv(file_path, sep=sep, decimal=decimal, header=int(header_row), usecols=[name_col, value_col]) # df.to_csv(f'{file_path}.df', sep=sep, decimal=decimal, index=True) # Filter the DataFrame to include only the measurands of interest df_filtered = df[df[name_col].isin(self.measurands)] df_filtered[value_col] = df_filtered[value_col].apply(convert_str_to_float) #print(df_filtered) if not scaling_factor == 1: print(f' scaling: {scaling_factor}') df_filtered[value_col] = df_filtered[value_col] * scaling_factor if not temperature == ref_temperature: print(f' temp. compensation: {temperature}, {ref_temperature}, {self.alpha}') df_filtered[value_col] = df_filtered[value_col].apply(self.TemperatureCorrection, args=(float(temperature), float(ref_temperature), float(self.alpha))) # print(df_filtered) combined_df_test = pd.concat([combined_df_test, df_filtered[value_col]], axis=1) # Append the file name to the list file_names.append(filename) #pd.set_option('display.max_rows', None) #pd.set_option('display.max_columns', None) #pd.set_option('display.width', None) #pd.set_option('display.max_colwidth', -1) #print(combined_df_test) combined_df_test.index = self.measurands combined_df_test.columns = file_names if ct_type == "values-sim": self.files_sim = file_names else: self.files_real = file_names combined_df_subtracted = combined_df_test.apply(lambda x: self.Deviation(x, ref_values), axis=1) print(combined_df_subtracted) os.makedirs(self.output_path, exist_ok=True) combined_df_test.to_csv(f"{self.output_path}/{self.name}_{ct_type}.csv", sep=self.sep, decimal=self.decimal, index=True) return combined_df_subtracted def TemperatureCorrection(self, Mvalue, Tmeas, Tcal, alpha): return Mvalue*(1-(alpha*(Tmeas-Tcal))) def Deviation(self, values, ref_values): #print(values.name) #print(ref_values) #print(values-ref_values[values.name]) return values - ref_values[values.name] def Edm_calc(self, RealValues, SimValues): T_real = float(self.metadata['values-real']['temperature']) T_ref = float(self.RealRefT) #print(SimValues) #print(RealValues) #print(SimValues) self.Real_avg = RealValues.mean(axis=1) self.Sim_avg = SimValues.mean(axis=1) u_psim = SimValues.std(axis=1) k_sim = get_value(self.metadata, ["values-sim", "k_value"], 2.0) u_cal = self.RealRefUncertainty u_p = RealValues.std(axis=1) u_ab = 0.2 * self.alpha u_b = (T_real - T_ref) * u_ab * self.Real_avg #print('u_b: ',u_b) k_real = get_value(self.metadata, ["values-real", "k_value"], 2.0) self.U_real = k_real * (u_cal.pow(2) + u_p.pow(2) + u_b.pow(2)).pow(1/2) self.U_sim = k_sim * u_psim self.E_DM = ((self.Sim_avg) - (self.Real_avg)) * (self.U_sim.pow(2) + self.U_real.pow(2)).pow(-0.5) #print(self.E_DM) self.Criterion_real = u_cal / u_p self.Criterion_sim = u_psim / u_p # self.Result = np.where((abs(self.E_DM) < 1) & (self.Criterion_real < 1) & (self.Criterion_sim < 1), True, False) self.Result = np.where((abs(self.E_DM) < 1) & (self.Criterion_real + self.Criterion_sim < 2), True, False) self.df["Real_avg"] = self.Real_avg self.df["u_cal"] = u_cal self.df["u_p"] = u_p self.df["u_b"] = u_b self.df["Real_U"] = self.U_real self.df["Sim_avg"] = self.Sim_avg self.df["Sim_U"] = self.U_sim self.df["E_DM"] = self.E_DM self.df["Criterion_real"] = self.Criterion_real self.df["Criterion_sim"] = self.Criterion_sim self.df["Criterion_sum"] = self.Criterion_real + self.Criterion_sim # self.df["Test Result"] = np.where((abs(self.df["E_DM"])< 1) & (self.df["Criterion_real"] < 1) & (self.df["Criterion_sim"] < 1), True, False) self.df["Test Result"] = np.where((abs(self.df["E_DM"])< 1) & (self.df["Criterion_real"] + self.df["Criterion_sim"] < 2), True, False) self.df.index.name = 'Measurand' # self.df.reset_index() print(self.df) # Save results. self.df.to_csv(f"{self.output_path}/{self.name}.csv", sep=self.sep, decimal=self.decimal, index=True, index_label='Measurand') def plotDeviations(self): import matplotlib.transforms from matplotlib.transforms import offset_copy fig = plt.figure(figsize=(7, 5)) # Add horizontal lines at 0 plt.axhline(y=0, color='lightgray') # plt.plot(self.df['Real_avg'], marker='o') plt.errorbar(self.df.index, self.df['Real_avg'], yerr=self.df['Real_U'], fmt='o') plt.errorbar([x+.15 for x in range(len(self.df.index))], self.df['Sim_avg'], yerr=self.df['Sim_U'], fmt='o') # Label the x-axis with names from self.measurands xtick_labels = [s[:7] for s in self.measurands] plt.xticks(ticks=range(len(xtick_labels)), labels=xtick_labels, rotation=90) # Add labels and title # plt.xlabel('Measurand') plt.ylabel('Deviation / mm') # plt.title('CTSimU2 Test Result') # plt.savefig(self.img_buf, format='png') # Show the plot #plt.show() plt.savefig(f"{self.output_path}/{self.name}_deviations.png") def plotResults(self): # assign categories and colormap cat = np.where(self.df['Criterion_sum'] < 2.0, 0, 1) col_map = np.array(['C0', 'C3']) # print(col_map[cat]) plt.figure(figsize=(10, 6)) # Add horizontal lines at -1 and +1 plt.axhline(y=-1, color='lightgray') plt.axhline(y=1, color='lightgray') plt.scatter(self.df.index, self.df["E_DM"], c=col_map[cat]) # Label the x-axis with names from self.measurands xtick_labels = [s[:7] for s in self.measurands] plt.xticks(ticks=range(len(xtick_labels)), labels=xtick_labels, rotation=90) # print(plt.xticks()) # Add labels and title # plt.xlabel('Measurand') plt.ylabel('$E_{DM}$ value') # plt.title('CTSimU2 Test Result') # plt.savefig(self.img_buf, format='png') # Show the plot #plt.show() plt.savefig(f"{self.output_path}/{self.name}_E_DM.png") def plotResult(self, wide = False): fig, (ax1, ax2) = plt.subplots(2, height_ratios=[2.25,1], sharex=True, dpi=120) ax1.set_title(self.name) # title = fig.suptitle(self.name) # title.set_position([.5,.9]) fig.subplots_adjust(hspace = 0.05) if wide: fig.set_size_inches(18, 6) max = 50 add = '_wide' fts = 'smaller' else: fig.set_size_inches(10, 6) max = 30 add = '' fts = None # Deviation plot # # Add horizontal lines at 0 ax1.axhline(y=0, color='lightgray') # Add the plot ax1.errorbar([s[:7] for s in self.df.index], self.df['Real_avg']*1e3, yerr=self.df['Real_U']*1e3, fmt='o', label='Real') ax1.errorbar([x+.15 for x in range(len(self.df.index))], self.df['Sim_avg']*1e3, yerr=self.df['Sim_U']*1e3, fmt='o', label='Sim') # ax1.set_ylim(-10,10) # Add labels and title ax1.set_axisbelow(False) ax1.tick_params(which='both', direction='in', top=True, right=True) # ax1.set_xlabel('Measurand') # ax1.title('CTSimU2 Test Result') ax1.set_ylabel('Deviation / µm') ax1.legend() # E_DM plot # # Add horizontal lines at -1 and +1 ax2.axhline(y=-1, color='lightgray', zorder=0) ax2.axhline(y=1, color='lightgray', zorder=0) # Add the plot passed = np.where((self.df["Criterion_real"] + self.df["Criterion_sim"] < 2), self.df["E_DM"], None) missed = np.where((self.df["Criterion_real"] + self.df["Criterion_sim"] < 2), None, self.df["E_DM"]) ax2.scatter(range(len(passed)), passed, label='passed exta criteria') ax2.scatter(range(len(missed)), missed, c='C3', label='missed exta criteria') ax2.set_ylim(-3,3) # Add labels and title ax2.set_axisbelow(False) ax2.tick_params(which='both', direction='in', top=True, right=True) # ax1.xlabel('Measurand') # ax1.title('CTSimU2 Test Result') ax2.set_ylabel('$E_{DM}$') # xtick_labels = [s[:7] for s in self.measurands] # ax2.xticks(ticks=range(len(xtick_labels)), labels=xtick_labels, rotation=90) ax2.legend() # X ticks and labels mul = 1 + int(len(self.df.index) / max) # print(mul) from matplotlib.ticker import (MultipleLocator, AutoMinorLocator) ax2.xaxis.set_major_locator(MultipleLocator(mul)) ax2.xaxis.set_minor_locator(MultipleLocator(1)) plt.xticks(fontsize=fts, ha='right', rotation=30) fig.savefig(self.img_buf, format='png') # Show the plot #plt.show() fig.savefig(f"{self.output_path}/{self.name}{add}.png") def TwinTest_report(self): from datetime import datetime from reportlab.lib.pagesizes import A4 from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle from reportlab.lib.units import mm from reportlab.platypus import Paragraph, PageBreak, Spacer, Image, Preformatted from ..report import Report, NumberedCanvas now = datetime.now() documentTitle = f"{self.output_path}/{self.name}.pdf" # Create a style sheet for the document styles = getSampleStyleSheet() # Define custom styles title_style = ParagraphStyle('Title', parent=styles['Heading1'], alignment=1, spaceBefore=10, spaceAfter=10) subtitle_style = ParagraphStyle('Subtitle', parent=styles['Heading2'], alignment=1, spaceBefore=6, spaceAfter=6) body_style = ParagraphStyle('Body', parent=styles['Normal'], fontSize=12, spaceBefore=0, spaceAfter=3) body_center_style = ParagraphStyle('Body', parent=styles['Normal'], alignment=1, fontSize=12, spaceBefore=0, spaceAfter=3) #print(body_style.parent.listAttrs()) # Define the content for the document title = Paragraph('Digital Model Test Report', title_style) #subtitle = Paragraph(self.name, subtitle_style) # creating a pdf object pdf = Report(documentTitle, pagesize = A4, showBoundary = 0, leftMargin = 27*mm, rightMargin = 20*mm, topMargin = 25*mm, bottomMargin = 25*mm, allowSplitting = 1, title = 'Digital Twin Test Report', creator = 'CTSimU Toolbox - https://github.com/BAMresearch/ctsimu-toolbox') pdf.setHeader('CTSimU Toolbox', 'Digital Model Test Report', f'{now:%Y-%m-%d}') pdf.setFooter(f'{self.name}.pdf', left2 = f'generated by CTSimU Toolbox {get_version()}', left3 = 'https://github.com/BAMresearch/ctsimu-toolbox') # Prepare the data for the table df = self.df.get(['Real_avg', 'Real_U', 'Sim_avg', 'Sim_U', "E_DM", 'Criterion_real', 'Criterion_sim', 'Test Result']).reset_index() df["Real_avg"] = df['Real_avg'].apply(lambda x: round(x, 4)) df["Real_U"] = df['Real_U'].apply(lambda x: round(x, 5)) df["Sim_avg"] = df['Sim_avg'].apply(lambda x: round(x, 4)) df["Sim_U"] = df['Sim_U'].apply(lambda x: round(x, 5)) df["E_DM"] = df["E_DM"].apply(lambda x: round(x, 2)) df['Criterion_real'] = df['Criterion_real'].apply(lambda x: round(x, 2)) df['Criterion_sim'] = df['Criterion_sim'].apply(lambda x: round(x, 2)) # Add the content to the PDF document elements = [title, # Paragraph("on", body_center_style), subtitle, Paragraph(f"generated at {now:%Y-%m-%d %H:%M}", body_center_style), Spacer(1, 24, True), # Paragraph(f"CTSimU Toolbox Version: {get_version()}", body_style), Paragraph(f'Metadata file: {self.current_metadata_file}', body_style), Paragraph(f"Description: {get_value(self.metadata, ['measurement','description'],'')}", body_style, bulletText='*'), Paragraph(f"Contact: {self.metadata['measurement']['contact']}", body_style, bulletText='*'), Paragraph(f"Experimental CTs: ", body_style, bulletText='*'), Paragraph(f"Description: {get_value(self.metadata, ['values-real','description'],'')}", body_style, bulletText=' -'), Paragraph(f"File path: {self.metadata['values-real']['csv_path']}", body_style, bulletText=' -'), Paragraph(f"Number of scans: {len(self.files_real)}", body_style, bulletText=' -'), Paragraph(f"Temperature: {self.metadata['values-real']['temperature']}", body_style, bulletText=' -'), Paragraph(f"Scaling: {self.metadata['values-real']['scaling_factor']}", body_style, bulletText=' -'), Paragraph(f"Reference: {self.metadata['reference-real']['description']}", body_style, bulletText=' -'), Paragraph(f"Simulated CTs: ", body_style, bulletText='*'), Paragraph(f"Description: {get_value(self.metadata, ['values-sim','description'],'')}", body_style, bulletText=' -'), Paragraph(f"File path: {self.metadata['values-sim']['csv_path']}", body_style, bulletText=' -'), Paragraph(f"Number of scans: {len(self.files_sim)}", body_style, bulletText=' -'), #Paragraph(f"Temperature: {self.metadata['values-sim']['temperature']}", body_style, bulletText=' -'), #Paragraph(f"Scaling: {self.metadata['values-sim']['scaling_factor']}", body_style, bulletText=' -'), Paragraph(f"Reference: {self.metadata['reference-sim']['description']}", body_style, bulletText=' -'), Image(f"{self.output_path}/{self.name}.png", 450, 270), # PageBreak(), # Paragraph('This is some content for the PDF document Page 2. ', body_style), pdf.df2table(df), Spacer(1, 24, True), Paragraph("Full imput metadata", subtitle_style), Paragraph(f'{self.current_metadata_file}:', body_style), Preformatted(json.dumps(self.metadata, indent=2), styles['Code']), Spacer(1, 24, True), Paragraph("List of output files", subtitle_style), Paragraph(f'{self.name}.csv', body_style), Paragraph(f'{self.name}.pdf (this document)', body_style), Paragraph(f'{self.name}.png', body_style), Paragraph(f'{self.name}_deviations.png', body_style), Paragraph(f'{self.name}_E_DM.png', body_style), Paragraph(f'{self.name}_values-real.csv', body_style), Paragraph(f'{self.name}_values-sim.csv', body_style), ] pdf.build(elements, canvasmaker=NumberedCanvas) def run(self): #self.RealValues = self.read_and_filter_csv_files(self.real_folder_path, "values-real") #print(self.RealValues) self.prepare_data() #self.SimValues = self.read_and_filter_csv_files(self.sim_folder_path, "values-sim") #print(self.SimValues) self.Edm_calc(self.RealValues, self.SimValues) self.plotDeviations() self.plotResults() self.plotResult() self.plotResult(wide = True) self.TwinTest_report()Methods
def Deviation(self, values, ref_values)def Edm_calc(self, RealValues, SimValues)def TemperatureCorrection(self, Mvalue, Tmeas, Tcal, alpha)def TwinTest_report(self)def plotDeviations(self)def plotResult(self, wide=False)def plotResults(self)def prepare_data(self)def read_and_filter_csv_files(self, folder_paths, ct_type)def read_metadata(self, filename: str = None)-
Import metadata from a CTSimU Twin Test file or a given metadata dictionary.
Parameters
filename:str-
Path to a CTSimU Twin Test file.
Default value:
None
def reset_metadata(self)-
Reset scenario's metadata information.
def run(self)