Reference kit for how to build end to end predictive assets maintenance pipeline.
Predictive assets maintenance is a proactive approach by leveraging historical data and effectively predicting potential failures to reduce costs, enhance safety and optimize asset performance. Compared with traditional inspection methods, predictive assets maintenance allows organizations to detect and address issues in early stages, preventing unexpected breakdowns and reducing equipment downtime. By identifying potential failures in advance, maintenance can be scheduled strategically to minimize disruptions and ensure continuous operation.
This workflow exactly demonstrates how to predict based on historical data and detect checking points (i.e. anomaly) through BigDL Time Series Toolkit (previous known as BigDL-Chronos or Chronos), therefore implement asset maintenance.
Check out more workflow examples here.
We provide an end-to-end (data loading, processing, built-in model, training, inference) AI solution with less operational cost by using BigDL Time Series Toolkit. The BigDL Time Series Toolkit supports building AI solution on single node or cluster and provides more than 10 built-in models for forecasting and anomaly detection. By deploying a model for predicting based on input data and another model for detecting the predicted data, we can effectively detect the potential abnormal behaviors or failures before any significant assets damage occurs.
This workflow shows how to use TSDataset to process time series data, TCNForecaster to predict future data, ThresholdDetector to detect anomalies and therefore implement predictive assets maintenance.
Specifically, users could load their historical time series data (datetime column is necessary) to TSDataset for preprocessing (e.g. impute, deduplicate, resample, scale/unscale, roll) and feature engineering (e.g. datetime feature, aggregation feature). And TSDataset can be initialized from pandas dataframe, path of parquet file and Prometheus data. It's recommended that users select proper data processing techniques based on characteristics of their dataset to implement better performance. The quality of input data will undoubtedly influence the results of training. Details about how to preprocess data can be found here.
The processed data could be fed into TCNForecaster for training and prediction. During initialization of TCNForecaster, users can specify the length of historical data, length of predicted data, number of variables to observe and number of variables to predict by setting parameter past_seq_len future_seq_len input_feature_num output_feature_num correspondingly.
Then input the predicted data and real data (real data is optional) to ThresholdDetector to generate anomaly time points. With these checking points, we can identify potential failures in advance and avoid asset loss in time.
The whole process is shown as following:
Provided example workflow shows how to implement predictive maintenance in the elevator industry. In detail, process history data of elevator system, predict the absolute value of vibration and detect the potential failure behaviors.
Users can refer to above documentation and this example to implement predictive assets maintenance with their customized datasets.
BigDL Time Series Toolkit and the workflow example shown below could be run widely on both Core™ and Xeon® series processers.
| Recommended Hardware | Precision | |
|---|---|---|
| CPU | Intel® 4th Gen Xeon® Scalable Performance processors | BF16 |
| CPU | Intel® 1st, 2nd, 3rd, and 4th Gen Xeon® Scalable Performance processors | FP32/INT8 |
| Recommended Hardware | Precision | |
|---|---|---|
| CPU | Intel® 4th Gen Xeon® Scalable Performance processors | BF16 |
| CPU | Intel® 1st, 2nd, 3rd, and 4th Gen Xeon® Scalable Performance processors | FP32/INT8 |
The hardware details of inference process is similiar to training.
Users first load their historical time series data (datetime column is necessary) to TSDataset for preprocessing and feature engineering. The processed data could be fed into TCNForecaster for training and prediction. Then input the predicted data and real data (real data is optional) to ThresholdDetector to generate anomaly time points. With these checking points, we can identify potential failures in advance and avoid asset loss in time.
The whole process is shown as following:
More details can be found in Solution Technical Details
Note: To realize satisfactory forecasting perfomance, users may need to select proper data processing techniques of TSDataset based on characteristics of their customized dataset. And parameters of TCNForecaster may also need to tuned for their customized datasets.
Define an environment variable that will store the workspace path, this can be an existing directory or one to be created in further steps. This ENVVAR will be used for all the commands executed using absolute paths.
export WORKSPACE=<your_workspace_dir>
Create a working directory for the workflow and all components of the workflow can be found here.
# For example...
mkdir -p $WORKSPACE && cd $WORKSPACE
git clone https://github.com/intel/Predictive-Assets-Maintenance.git
cd $WORKSPACE/Predictive-Assets-Maintenance/AIOps
The dataset we will use in our workflow example is Elevator Predictive Maintenance Dataset, which is not contained in our docker image or workflow directory. The dataset contains operation data in the elevator industry, including lots of electromechanical sensors, ambiance and physics data.
Before runnning the workflow script, you need to download dataset from kaggle and decompress it to get csv file called predictive-maintenance-dataset.csv.
export DATASET_DIR=<your_directory_of_csv_file>
cd $DATASET_DIR
# download and decompress to get csv file called `predictive-maintenance-dataset.csv`
cd $WORKSPACEYou can execute the references pipelines using the following environments:
- Docker
- Argo
- Bare metal.
- Jupyter lab.
Follow these instructions to set up and run our provided Docker image. For running on bare metal, see the bare metal instructions.
You'll need to install Docker Engine on your development system. Note that while Docker Engine is free to use, Docker Desktop may require you to purchase a license. See the Docker Engine Server installation instructions for details.
Ensure you have Docker Compose installed on your machine. If you don't have this tool installed, consult the official Docker Compose installation documentation.
DOCKER_CONFIG=${DOCKER_CONFIG:-$HOME/.docker}
mkdir -p $DOCKER_CONFIG/cli-plugins
curl -SL https://github.com/docker/compose/releases/download/v2.7.0/docker-compose-linux-x86_64 -o $DOCKER_CONFIG/cli-plugins/docker-compose
chmod +x $DOCKER_CONFIG/cli-plugins/docker-compose
docker compose versionEnsure some environment variables are set before running workflow.
export DATASET_DIR=<your_directory_of_csv_file>
export FINAL_IMAGE_NAME=workflow # You may choose any image name you want
export http_proxy=<your_http_proxy>
export https_proxy=<your_https_proxy>Then build the provided docker image.
cd $WORKSPACE/Predictive-Assets-Maintenance/docker-compose
docker compose up --buildThe container flow diagram using Mermaid is as following:
Run entire pipeline:
docker compose run asset-maintenance &| Environment Variable Name | Default Value | Description |
|---|---|---|
| DATASET_DIR | $DATASET_DIR |
Preprocessed Dataset |
Follow logs of each individual pipeline step using the commands below:
docker compose logs asset-maintenance -fThe Mermaid diagram for workflow is shown as following:
Create your own script and run your changes inside of the container using compose as follows:
cd $WORKSPACE/Predictive-Assets-Maintenance/docker-compose
docker compose run dev| Environment Variable Name | Default Value | Description |
|---|---|---|
| DATASET_DIR | $DATASET_DIR |
Preprocessed Dataset |
| SCRIPT | src/own-script.sh |
Name of Script |
If your environment requires a proxy to access the internet, export your development system's proxy settings to the docker environment:
export DOCKER_RUN_ENVS="-e ftp_proxy=${ftp_proxy} \
-e FTP_PROXY=${FTP_PROXY} -e http_proxy=${http_proxy} \
-e HTTP_PROXY=${HTTP_PROXY} -e https_proxy=${https_proxy} \
-e HTTPS_PROXY=${HTTPS_PROXY} -e no_proxy=${no_proxy} \
-e NO_PROXY=${NO_PROXY} -e socks_proxy=${socks_proxy} \
-e SOCKS_PROXY=${SOCKS_PROXY}"Run the workflow using the docker run command, as shown:
export DATASET_DIR=<your_directory_of_csv_file>
docker run \
--env DATASET_DIR=${DATASET_DIR} \
--volume ${DATASET_DIR}:/dataset \
--volume ${PWD}:/workspace \
--workdir /workspace \
--privileged --init -it --rm \
<workflow image name>Stop containers created by docker compose and remove them.
docker compose downIf you don't have this tool installed, consult the official Installing Helm.
curl -fsSL -o get_helm.sh https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 && \
chmod 700 get_helm.sh && \
./get_helm.shBefore running workflow on K8s, we need to set up docker image and working directory (i.e. hostpath).
cd $WORKSPACE/<workflow repo name>/helm-charts/docker
bash build-docker-image.sh # NAME:TAG of the image is intelanalytics/bigdl:asset-maintenance-ubuntu-20.04- Install Argo Workflows and Argo CLI
- Configure your Artifact Repository
- Ensure that your dataset and config files are present in your chosen artifact repository.
Prepare and update workflow values in values.yml
cd $WORKSPACE/<workflow repo name>/helm-charts/kubernetes
# update workflow values in `values.yml`Then run this workflow on K8s using helm.
export NAMESPACE=argo
cd helm-charts/kubernetes
helm install asset-maintenance ./
argo submit --from wftmpl/<workflow-name> --namespace=${NAMESPACE}To view your workflow progress:
argo logs @latest -fFollow these instructions to set up and run this workflow on your own development system. For running a provided Docker image with Docker, see the Docker instructions.
Our examples use the conda package and enviroment on your local computer.
If you don't already have conda installed, see the Conda Linux installation
instructions.
Run these commands to set up the workflow's conda environment and install required software:
cd $WORKSPACE
git clone https://github.com/intel/Predictive-Assets-Maintenance.git
cd $WORKSPACE/Predictive-Assets-Maintenance/AIOps
conda create -n my_env python=3.9 setuptools=58.0.4
conda activate my_env
pip install bigdl-chronos[pytorch]==2.4.0b20230522 matplotlib==3.7.1 notebook==6.4.12
Besides, download dataset from kaggle and decompress it to get csv file called predictive-maintenance-dataset.csv.
Run the following command to download and extract dataset.
export DATASET_DIR=<your_directory_of_csv_file>Now we can use these commands to run the workflow:
# transform the notebook to a runnable python script
jupyter nbconvert --to python AIOps_Asset_Maintenance.ipynb
sed -i '/get_ipython()/d' AIOps_Asset_Maintenance.py
sed -i "s#/dataset/predictive-maintenance-dataset.csv#$DATASET_DIR/predictive-maintenance-dataset.csv#g" AIOps_Asset_Maintenance.py
python AIOps_Asset_Maintenance.pyThis workflow provides instructions on how to implement asset maintenance through a Temporal Convolution Neural Network and a Threshold Detector on BigDL Time Series Toolkit using time series dataset as an example.
The main part will be the model training progress bar, finally you can obtain the checking points which exist potential issues and need to pay attention in test dataset.
Epoch 4: 100%|███████████████████████████████████████████████████████████████████████| 1814/1814 [01:39<00:00, 18.32it/s, loss=0.311]
Training completed
MSE is 25.46
The index of anomalies in test dataset only according to predict data is:
pattern anomaly index: []
trend anomaly index: [3199, 3159, 3079]
anomaly index: [3079, 3159, 3199]
The index of anomalies in test dataset according to true data and predicted data is:
pattern anomaly index: [279, 679, 1079, 1479, 1879, 2279, 2319, 2679, 2719, 2739, 2751, 2783, 3079, 3119, 3139, 3151, 3183, 3479, 3519, 3551, 3879, 3919, 4279, 4679, 5079, 5479, 5879, 6279]
trend anomaly index: [2959, 3599, 2839, 2719, 3359, 2599, 3239, 3119, 2999, 3639, 2879, 3519, 2759, 3399, 2639, 3279, 3159, 2519, 3039, 3199, 2919, 3559, 2799, 3439, 3319, 2559]
anomaly index: [3079, 6279, 5879, 2319, 2959, 3599, 3479, 279, 2839, 2719, 3359, 679, 3879, 2599, 3239, 3119, 2739, 1079, 4279, 2999, 3639, 3319, 3519, 2751, 2879, 3139, 1479, 4679, 2759, 3399, 2559, 3151, 3919, 2639, 3279, 1879, 5079, 3159, 2519, 2783, 3551, 3039, 2279, 5479, 2919, 3559, 3183, 2799, 3439, 2679, 3199]
Users are encouraged to use the conda package and enviroment on local computer. If you don't already have conda installed, see the Conda Linux installation instructions.
conda create -n my_env python=3.9 setuptools=58.0.4
conda activate my_env
pip install --pre --upgrade bigdl-chronos[pytorch] matplotlib notebook==6.4.12
# to use jupyter lab
pip install jupyterlab jupyter_server_terminals
conda install ipykernel -y
ipython kernel install --user --name=my_env# If you have done this in section Workflow Repository
# please skip next 2 commands
mkdir ~/work && cd ~/work
git clone https://github.com/intel/Predictive-Assets-Maintenance.gitRun the following command to use jupyter lab:
jupyter labOpen jupyter lab in a web browser and set my_env as the jupyter kernel. Then you can open AIOps_Asset_Maintenance.ipynb and follow the notebook's instructions step by step.
The workflow example shows how to use BigDL Time Series Toolkit to implement predictive assets maintenance. Users may refer to example notebook to obtain more details.
For more information about or to read about other relevant workflow examples, see these guides and software resources:
- Document First page (you may start from here and navigate to other pages)
- How to Guides (If you are meeting with some specific problems during the usage, how-to guides are good place to be checked.)
- Example & Use-case library (Examples provides short, high quality use case that users can emulated in their own works.)
Nothing for now.
If you have any questions with this workflow, the team tracks both bugs and enhancement requests using GitHub issues.
These materials are intended to assist designers who are developing applications within their scope. These materials do not purport to provide all of the requirements for a commercial, productions, or other solution. Any commercial or productions use of solution based on or derived from these materials is beyond their scope. You are solely responsible for the engineering, testing, safety, qualification, validation, and applicable approvals for any solution you build or use based on these materials. Intel bears no responsibility or liability for such use. You are solely responsible for using your independent analysis, evaluation and judgment in designing your applications and have full and exclusive responsibility to assure the safety of your applications and compliance of your applications with all applicable regulations, laws and other applicable requirements. You further understand that you are solely responsible for obtaining any licenses to third-party intellectual property rights that may be necessary for your applications or the use of these materials.
To the extent that any public or non-Intel datasets or models are referenced by or accessed using these materials those datasets or models are provided by the third party indicated as the content source. Intel does not create the content and does not warrant its accuracy or quality. By accessing the public content, or using materials trained on or with such content, you agree to the terms associated with that content and that your use complies with the applicable license.
Intel expressly disclaims the accuracy, adequacy, or completeness of any such public content, and is not liable for any errors, omissions, or defects in the content, or for any reliance on the content. Intel is not liable for any liability or damages relating to your use of public content.
Intel’s provision of these resources does not expand or otherwise alter Intel’s applicable published warranties or warranty disclaimers for Intel products or solutions, and no additional obligations, indemnifications, or liabilities arise from Intel providing such resources. Intel reserves the right, without notice, to make corrections, enhancements, improvements, and other changes to its materials.
Intel technologies may require enabled hardware, software or service activation. Performance varies by use, configuration and other factors. No product or component can be absolutely secure.
Intel is committed to respecting human rights and avoiding complicity in human rights abuses. See Intel's Global Human Rights Principles. Intel's content is intended only to be used in applications that do not cause or contribute to a violation of an internationally recognized human right.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.