Energy Management Using Real-Time Non-Intrusive Load Monitoring

Introduction

The goal of non-intrusive load monitoring (NILM) is to recover the energy consumption of individual appliances from the aggregate mains signal, which is a measure of the total electricity consumption of a building or house. NILM is also known as energy disaggregation and both terms will be used interchangeably throughout.

R. Gopinath, et al., nicely summarizes the rational behind NILM in the paper Energy management using non-intrusive load monitoring techniques — State- of-the-art and future research directions¹:

In recent years, the development of smart sustainable cities has become the primary focus among urban planners and policy makers to make responsible use of resources, conserve the environment and improve the well-being of the society. Energy management is an integral part of the smart sustainable cities development programme which involves conscious and efficient use of available energy resources towards attaining sustainability and self-reliance on energy systems. Building sector is one of the key sectors that utilize more energy. Therefore, efforts are being made to monitor and manage energy consumption effectively in residential and commercial buildings. In recent years, non-intrusive load monitoring (NILM) technique has become a popular and emerging approach to monitor events (on/off) and energy consumption of appliances/electrical utilities in buildings using single energy meter. The information about the energy consumption at the appliance level would help consumers to understand their appliance usage behavior and take necessary steps for reducing energy consumption.

By using deep learning models trained on publicly available datasets it is now feasible to enable NILM in very cost effective ways by leveraging commodity hardware and open source software. Deploying the trained models at the edge of the power grid, i.e. the building level, results in additional cost savings since it obviates the need for always connected internet cloud services which can be expensive at scale and running the algorithms in real-time provides for low latency operation.

This Project will show you how NILM works by taking you through the steps I used to implement a prototype system at my house.

Architecture

The following first two diagrams illustrate the NILM concept and process steps at a high level. The last diagram shows my prototype system which is based on these concepts and is implemented using Arduino- and Raspberry Pi-based compute.

NILM Algorithm Selection and Models

Algorithm Selection

Energy disaggregation is a highly underdetermined and a single-channel Blind Source Separation (BSS) problem² which makes it difficult to obtain accurate predictions. Let $M$ be the number of household appliances and $i$ be the index referring to the $i$-th appliance. The aggregate power consumption $x$ at a given time $t$ is the sum of the power consumption of all appliances $M$, denoted by $y_i\forall{i=1,...,M}$. Therefore, the total power consumption $x$ at a given time $t$ can expressed by the equation below.

$$ x(t)=\sum_{i=1}^{M} y_i(t) + \epsilon_{\text{noise}}(t) $$

(1)

Where $\epsilon_{noise}$ is a noise term. The goal of this project is to solve the inverse problem and estimate the appliance power consumption $y_i$, given the aggregate power signal $x$, and to do so in a manner suitable for deployment at the edge.

Past approaches have included factorial hidden Markov models (FHMM)¹ and various event-based methods with some success⁶. You can also solve the single-channel BSS problem can by using sequence-to-sequence (seq2seq) learning with neural networks, and it can applied to the NILM problem using transformers, convolutional and recurrent neural networks³. Seq2seq learning involves training a neural network to map between an input time series, such as the aggregate power readings the case of NILM, and a output sequence, such as the estimated energy consumption of a single appliance. A sliding input window is typically used to training the network which generates a corresponding window of the output. This method produces multiple predictions for each appliance in the output so an average of the predictions is used for the final result. Some of the predictions will be more accurate than others, especially those near the midpoint of the input sliding window. The averaging will tend to lower the overall accuracy of the predictions.

Some of the disadvantages of of seq2seq leaning can mitigated by sequence-to-point learning (seq2point) for single-channel BSS⁴. You also use a sliding input signal window in this approach, however the network is trained to predict the output signal only at the midpoint of the window which makes the prediction problem easier on the network, leading to more accurate results.

I selected the seq2point learning approach for my prototype system and my implementation was inspired and guided by work described by Michele D'Incecco, Stefano Squartini and Mingjun Zhong⁵. I developed a variety of seq2point learning models using Tensorflow as shown in the Python module define_models.py but focussed my work on the models transformer and cnn.

Neural Network Models

The cnn model is depicted below for an input sequence length of 599 samples. The model generally follows traditional cnn concepts from vision use cases where several convolutional layers are used to extract features from the input power sequence at gradually finer details as the input traverses through the network. These features are the on-off patterns of the appliances as well as their power consumption levels. Max pooling is used to manage the complexity of the model after each convolutional layer. Finally, dense layers are used to output the final single point power consumption estimate for the window which is de-normalized before using in downstream processing. There are about 40 million parameters in this model using the default values.

The transformer model is depicted below for an input sequence length of 599 samples where the transformer block is a Bert-style encoder. The input sequence is first passed through a convolutional layer to expand it into a latent space which is analogous to the feature extraction in the cnn model case. These features are pooled and L2 normalized to reduce model complexity and to mitigate the effects of outliers. Next, the sequence features are processed by a Bert-style transformer lineup which includes positional embedding and transformer blocks that applies importance weighting. The output of the encoder is decoded by several layers which are comprised of relative position embedding which applies symmetric weights around the mid-point of the signal, average pooling which reduces the sequence to a single value per feature and then finally dense layers that output the final single point estimated power value for the window which again is de-normalized for downstream processing. There are about six million parameters in this model using the default values.

The Bert-style transformer encoder is depicted below.

NILM Datasets

There are a number of large-scale publicly available datasets specifically designed to address the NILM problem which were captured in household buildings from various countries. The table⁷ below shows several of the most widely used.

The datasets generally include many 10’s of millions of active power, reactive power, current, and voltage samples but with different sampling frequencies which requires you to pre-process the data before use. Most NILM algorithms utilize only real (aka active or true) power data. Five appliances are usually considered for energy disaggregation research which are kettle, microwave, fridge, dish washer and washing machine. These are the appliances I considered for my prototype and following the work of Michele DIncecco, et al.⁵, I mainly focused on the REFIT⁸ dataset but will eventually include UK-DALE and REDD.

Note that these datasets are typically very imbalanced because the majority of the time an appliance is in the off state.

Model Training and Results

I used TensorFlow to train and test the model. All code associated with this section can be found on the Machine Learning section of the project’s GitHub, NILM⁹. The seq2point learning models for the appliances were trained individually on z-score standardized REFIT data or normalized to $[0, P_m]$, where $P_m$ is the maximum power consumption of an appliance in its active state. Using normalized data tended to give the best model performance so it is used by default.

I used the following metrics to evaluate the model’s performance. You can view the code that calculates these metrics here.

• Mean absolute error ($MAE$), which evaluates the absolute difference between the prediction and the ground truth power at every time point and calculates the mean value, as defined by the equation below.

$$ \mathrm{MAE}=\frac{1}{N}\sum_{i=1}^{N}\lvert \hat{x}_i - x_i\rvert $$

(2)

• Normalized signal aggregate error ($SAE$), which indicates the relative error of the total energy. Denote $r$ as the total energy consumption of the appliance and $\hat{r}$ as the predicted total energy, then SAE is defined per the equation below.

$$ \mathrm{SAE}=\fr