node-red-contrib-self-healing/kalman-noise-filter/kalman.html

SHEN: Self-healing extensions for Node-RED.

<script type="text/javascript">
  RED.nodes.registerType("kalman-noise-filter", {
    category: "SHEN",
    color: "#74b9ff",
    defaults: {
      r: { value: 0.01, required: true, validate: RED.validators.number() },
      q: { value: 3, required: true, validate: RED.validators.number() },
    },
    inputs: 1,
    outputs: 1,
    icon: "status.svg",
    label: function () {
      return this.name || "kalman-filter";
    },
  });
</script>

<script type="text/html" data-template-name="kalman-noise-filter">
  <div class="form-row">
    <label for="node-input-r">
      <i class="icon-tag"></i> R (Process Noise)
    </label>
    <input type="text" id="node-input-r" placeholder="0.01" />
  </div>
  <div class="form-row">
    <label for="node-input-q">
      <i class="icon-tag"></i> Q (Measurement Noise)
    </label>
    <input type="text" id="node-input-q" placeholder="3" />
  </div>
</script>

<script type="text/html" data-help-name="kalman-noise-filter">
  <p>Apply a Kalman Noise Filter to numerical message payloads</p>
  <p>
    It uses previous and current measurements and statistics to predict the next
    value.
  </p>
  <p>
    It assumes that the data is mostly constant, and that the noise is Gaussian.
  </p>
  <h3>Inputs</h3>
  <dl class="message-properties">
    <dt>
      payload <span class="property-type">number | array&ltnumber&gt</span>
    </dt>
    <dd>the data you want to filter</dd>
  </dl>
  <h3>Outputs</h3>
  <dl class="message-properties">
    <dt>
      payload <span class="property-type">number | array&ltnumber&gt</span>
    </dt>
    <dd>filtered data</dd>
  </dl>
  <h3>Details</h3>
  <p>
    This node takes the inputs given to it either as a number or an array of
    numbers and runs them through a thing called a Kalman filter. It is a
    statistical predictor that is used to reduce the effect of random noise on
    measurements.
  </p>
  <p>
    It is assumed that the system is unidimensional, in a more or less constant
    state, and subjected to both internal noise and measurement noise (both
    gaussian). Therefore it is configurable in two parameters:
  </p>
  <ul>
    <li>
      Process Noise (R): noise that is internal to the system. e.g. human body
      temperature, it might not be exactly 37C all the time, but a bit above or
      below it.
    </li>
    <li>Measurement Noise (Q): the noise introduced by the measurement.</li>
  </ul>
  <p>
    The node itself functions in a simple way. It is initialized with its own
    filter, and sequentially runs the successive values through it, returning
    the estimations.
  </p>
  <h3>References</h3>
  <ul>
    <li>
      <a href="https://en.wikipedia.org/wiki/Kalman_filter">Kalman filter</a> on
      Wikipedia
    </li>
    <li>
      <a href="https://github.com/wouterbulten/kalmanjs">Implementation used</a>
    </li>
  </ul>
</script>