pymodal 0.2.1

Introduction

This library is a work in progress developed by the Applied Mechanics Lab, IQS School of Engineering, as part of the research in Structural Health Monitoring (SHM). This library enables the user to store and process Frequency Response Functions (FRF). The library includes a wide range of frequency-based damage indicators found in the literature. Currently, Transmissibility Functions (TF) are being implemented in the library. The library also includes tools for building certain geometries in ANSYS and get their FRFs. If you plan on using the ANSYS module, be sure to have a working ANSYS installation.

Installation

In order to install this module, just run

pip install pymodal

in your terminal. This will also potentially install all the requirements, which you can find in requirements.txt, although they will be included here as well for clarity’s sake:

• numpy

• scipy

• matplotlib

• pandas

• pyansys

pip install -e .[dev]

Basic usage

Make an instance of FRF class:

At least one of the following must be specified: Resolution, Bandwidth or Maximum frequency. If not specified, minimum frequency is assumed to be 0 Hz.

frf_data = pymodal.FRF(
    frf,
    resolution,
    bandwidth,
    max_freq,
    min_freq,
    name,
    part,
    modal_frequencies
    )

Generate instance named frf_data from frf numpy array.

Add new FRFs to existing instance:

def extend(self, frf: list, name: list = None)

Save instance as zip file:

def extend(self, frf: list, name: list = None)

Plot FRF: Plot all FRFs together with varying colors unless otherwise specified.

def plot(self,
         ax: list = None,
         fontsize: float = 12,
         title: str = 'Frequency Response',
         title_size: float = None,
         major_locator: int = 4,
         minor_locator: int = 4,
         fontname: str = 'serif',
         color: list = None,
         ylabel: str = None,
         bottom_ylim: float = None,
         decimals_y: int = 1,
         decimals_x: int = 1):

Use slice to only plot specific FRFs.

frf_data[0].plot()
plt.show()

Post-processing FRFs

Change resolution

frf.change_resolution(frequencies=[0,100])

Change FRF lines

Change frequency range

Extract real part of FRF

Extract imaginary part of FRF

Calculate magnitude of FRF

Calculate phase of FRF

Extract modal frequencies

Extract mode shapes

Synthetic FRFs

Silhouette

Transmissibility matrix

Frequency-based damage indicators

Currently, the pymodal library holds the following damage indicators:

Frequency Response Function RMS [FRFRMS]:

def get_FRFRMS(self, ref:int)

https://www.sciencedirect.com/science/article/abs/pii/S1270963802011938

Global Amplitude Criterion [GAC]:

def get_GAC(self, ref:int, frf: list = None)

Average Integration Global Amplitude Criterion (AIGAC):

def get_AIGAC(self, ref:int)

Frequency Domain Assurance Criterion (FDAC):

Response Vector Assurance Criterion (RVAC)

Detection and Relative Quantification (DRQ)

Detection and Relative Quantification curvature-based (DRQ’’)

Frequency Response Function Scale Factor (FRFSF)

Coefficient of Determination (R^2)

ODS difference indicator (∆ODS)

Frequency Response Function Similarity Metric (FRFSM)

Complex Frequency Domain Assurance Crietrion [CFDAC]

Spectral Correlation Index [SCI] https://www.sciencedirect.com/science/article/abs/pii/S0888327018306551?via%3Dihub

def get_SCI(self, ref:int, part: str = 'abs')

MCP server

Pymodal ships an MCP server that exposes collection construction, signal processing, FRF computation, SHM-indicator math and PyTorch-dataset hand-off as tools an LLM agent can call directly.

Install the optional dependency:

pip install -e .[mcp]

Run the server (stdio transport):

python -m pymodal.mcp
# or, after install:
pymodal-mcp

Configure your MCP client (Claude Desktop, Claude Code, …) to launch python -m pymodal.mcp as a server. Every tool takes file paths in and writes file paths out, so a typical agent flow is

create_timeseries_collection → add_gaussian_noise → timeseries_to_frf → compute_indicator → split_collection → torch_dataset_summary.

Use list_indicators and list_collection_classes for in-tool discovery.

ANSYS