Control engine system and method
View Patent ↗The present disclosure provides a control engine system connected to a BAS, the control engine system includes a data preprocessing engine, a random forest generator, a tree selector, and a BAS code generator. The data preprocessing engine receives a data from the BAS and is configured for performing preprocessing on the data and generating a training data accordingly. The random forest generator is connected to the data preprocessing engine, the random forest generator receives the training data and generates a random forest accordingly. The tree selector is connected to the random forest generator, and the tree selector receives the random forest for selecting a final decision tree. The BAS code generator is connected to the tree selector, the BAS code generator receives the final decision tree and encodes a supervisory decision tree according to the final decision tree, the BAS code generator outputs the supervisory decision tree to the BAS.
1 . A control engine system connected to a building automation system (BAS), comprising:
a data preprocessing engine, wherein the data preprocessing engine is adapted to receive a data from the BAS and configured for performing preprocessing on the data and generating a training data accordingly;
a random forest generator, connected to the data preprocessing engine, wherein the random forest generator is configured to receive the training data and generate a random forest according to the training data;
a tree selector, connected to the random forest generator, wherein the tree selector is configured to receive the random forest for selecting a final decision tree; and
a BAS code generator, connected to the tree selector, wherein the BAS code generator is configured to receive the final decision tree and encode a supervisory decision tree according to the final decision tree, and the BAS code generator is configured to output the supervisory decision tree to the BAS, wherein the supervisory decision tree is applied to the BAS to save energy and/or improve comfort.
2 . The control engine system according to claim 1 , wherein the tree selector comprises a criteria-based indicative tree selector.
3 . The control engine system according to claim 1 , wherein the data from the BAS comprises at least one of a control structure and a thermal sensor data.
4 . The control engine system according to claim 1 , wherein the preprocessing of the data preprocessing engine comprises at least one of the normalizing, slicing, and shuffling of the data.
5 . The control engine system according to claim 1 , wherein the random forest generator is configured to generate the random forest using a classification and regression tree (CART) algorithm.
6 . The control engine system according to claim 1 , wherein the control engine system comprises an agnostic control engine (ACE).
7 . A control engine method, applicable for a control engine system, wherein the control engine system is connected to a BAS and comprises a data preprocessing engine, a random forest generator, a tree selector and a BAS code generator, and the control engine method comprises steps of:
(a) obtaining a data from a BAS and determining a feature data corresponding to the data;
(b) receiving the feature data corresponding to the data;
(c) performing a preprocessing to the feature data by the data preprocessing engine;
(d) generating a random forest corresponding to the BAS using the feature data by the random forest generator;
(e) selecting a final decision tree according to the random forest by the tree selector; and
(f) generating a supervisory decision tree according to the final decision tree by the BAS code generator, wherein the supervisory decision tree is applied to the BAS to save energy and/or improve comfort.
8 . The control engine method according to claim 7 , wherein the feature data comprises at least of the states, inputs, disturbances or outputs of the BAS.
9 . The control engine method according to claim 7 , wherein the random forest is statistically significant and/or of a default number of trees.
10 . The control engine method according to claim 7 , further comprising a step of: sorting the random forest by a node count to produce the final decision tree.
11 . A control engine method for a model-based predictive control (MPC) configuration, comprising the steps of the control engine method of claim 7 , and further comprising steps of:
(g) receiving preprocess data from the BAS;
(h) fitting a MPC model to the preprocess data and tunning a MPC parameter of the MPC model for a desired optimal performance of a control object;
(i) writing the MPC model into the BAS to override an original control of the BAS; and
(j) evaluating the performance of the MPC model.
12 . The control engine method according to claim 11 , wherein the preprocess data is in time series.
13 . The control engine method according to claim 11 , wherein in the step (h), the fitting of the MPC model is performed by using least squares fitting, and the MPC model comprises a linear time-invariant multi-input model.
14 . The control engine method according to claim 11 , wherein in the step (h), the fitting of the MPC model is performed by using a nonlinear function of parameters from a state-space model.
15 . The control engine method according to claim 11 , wherein the evaluation of the performance of the MPC model comprises steps of: estimating a potential performance improvement or a periodic evaluation of the MPC model performance comparing to estimates, and re-fitting or re-training the MPC model.