Predictive maintenance system for building equipment with reliability modeling based on natural language processing of warranty claim data
A method for generating a training data set includes receiving, by a processing circuit, warranty claim data associated with one or more building devices or building device components; processing, by the processing circuit, the warranty claim data using natural language processing to generate a training data set comprising one or more causes and solutions associated with failure of the one or more building devices or the building device components; and training, by the processing circuit, a component reliability model using the training data set to produce a trained model.
1 . A method for generating a reliability model for one or more building devices or building device components, the method comprising:
receiving, by a processing circuit, warranty claim data associated with the one or more building devices or the building device components;
processing, by the processing circuit, the warranty claim data using natural language processing to generate a training data set by:
generating a plurality of clusters of warranty claims, each cluster based on a noun representing one of the building device components identified in the warranty claim as part of both one or more causes and one or more solutions associated with failure of the one or more building devices or the building device components; and
selecting, for the training set, a most frequent cause and solution of the one or more causes and the one or more solutions from a quantity of the plurality of clusters;
training, by the processing circuit, a component reliability model using the training data set to produce a trained model, wherein training the component reliability model comprises configuring the component reliability model to output a reliability metric related to the one or more causes by adjusting parameters of the component reliability model; and
in response to predicting a failure of a building device or component of the one or more building devices or building device components using the component reliability model, at least one of:
generating, by the processing circuit, electronic control signals for communicating a shutdown command to the building device or component and transmitting, by the processing circuit, the electronic control signals to the building device or component, causing the building device or component to cease current operations; or
generating, by the processing circuit, electronic control signals for communicating operations based on the current operations of the building device or component to an alternative building device or component and transmitting, by the processing circuit, the electronic control signals to the alternative building device or component, causing the alternative building device or component to perform the operations.
2 . The method of claim 1 , wherein the warranty claim data includes a warranty claim comment, wherein processing the warranty claim data comprises at least one of identifying key words in the warranty claim comment, removing a stop word in the warranty claim comment, lemmatizing words in the warranty claim comment, or removing unnecessary words from the warranty claim comment.
3 . The method of claim 1 , wherein evaluating the processed warranty claim data using natural language processing comprises:
identifying, by the processing circuit, a plurality of words which occur in the warranty claim data over a predetermined amount of time;
determining, by the processing circuit, that the plurality of words are independent from each other;
creating, by the processing circuit, a plurality of word clusters associated with each of the plurality of words that are independent from each other; and
generating, by the processing circuit, a plurality of n-grams for the plurality of word clusters to determine the one or more causes and the one or more solutions associated with the failure of the one or more building devices or the building device components.
4 . The method of claim 3 , wherein generating the training data set based on the one or more causes and the one or more solutions further comprises combining, by the processing circuit, the plurality of n-grams to form the training data set which describes a predetermined number of building component failure causes and solutions.
5 . The method of claim 1 , further comprising:
generating, by the processing circuit, the reliability metric describing a predicted failure time associated with the one or more building devices or the building device components based on the trained model; and
updating, by the processing circuit, the building device components based on the reliability metric.
6 . The method of claim 5 , wherein updating the building device components comprises automatically updating software for the building device components.
7 . The method of claim 1 , wherein training the component reliability model to produce the trained model includes determining a shape parameter and a scale parameter of a Weibull model.
8 . The method of claim 1 , wherein processing the warranty claim data includes filtering out unnecessary information and identifying key words.
9 . One or more non-transitory computer-readable storage media having instructions stored thereon that, when executed by one or more processors, cause the one or more processors to:
receive warranty claim data associated with one or more building devices or building device components;
process the warranty claim data using natural language processing to generate a training data set by:
generating a plurality of clusters of warranty claims, each cluster based on a noun representing one of the building device components identified in the warranty claim as part of both one or more causes and one or more solutions associated with failure of the one or more building devices or the building device components; and
selecting, for the training set, a most frequent cause and solution of the one or more causes and the one or more solutions from a quantity of the plurality of clusters;
train a component reliability model using the training data set to produce a trained model, wherein training the component reliability model comprises configuring the component reliability model to output a reliability metric related to the one or more causes by adjusting parameters of the component reliability model; and
in response to predicting a failure of a building device or component of the one or more building devices or building device components using the component reliability model, at least one of:
generate electronic signals for communicating (i) an identification of the building device or component to a work order management system and (ii) a command to generate an automated work order request over a network and transmit the electronic signals on the network to the work order management system, causing the work order management system to generate the automated work order request including the identification of the building device or component; or
generate electronic signals for communicating (i) an identification of a part of the building device or component to a part ordering system or parts supplier system and (ii) a request to order the part for the building device or component to generate to the part ordering system or the parts supplier system and transmit the electronic signals on the network to the part ordering system or the parts supplier system, causing the part to be ordered.
10 . The one or more non-transitory computer-readable storage media of claim 9 , wherein the warranty claim data includes a warranty claim comment.
11 . The one or more non-transitory computer-readable storage media of claim 10 , wherein processing the warranty claim data comprises at least one of identifying key words in the warranty claim comment, removing a stop word in the warranty claim comment, lemmatizing words in the warranty claim comment, or removing unnecessary words from the warranty claim comment.
12 . The one or more non-transitory computer-readable storage media of claim 9 , wherein evaluating the processed warranty claim data using the natural language processing comprises:
identifying a plurality of words which occur in the warranty claim data over a predetermined amount of time;
determining that the plurality of words are independent from each other;
creating a plurality of word clusters associated with each of the plurality of words that are independent from each other; and
generating a plurality of n-grams for the plurality of word clusters to determine the one or more causes and the one or more solutions associated with the failure of the one or more building devices or the building device components.
13 . The one or more non-transitory computer-readable storage media of claim 12 , wherein generating the training data set based on the one or more causes and the one or more solutions further comprises combining the plurality of n-grams to form the training data set which describes a predetermined number of building component failure causes and solutions.
14 . The one or more non-transitory computer-readable storage media of claim 9 , wherein the instructions further cause the one or more processors to:
generate the reliability metric describing a predicted failure time associated with the one or more building devices or the building device components based on the trained model; and
update the building device components based on the reliability metric.
15 . The one or more non-transitory computer-readable storage media of claim 14 , wherein updating the building device components comprises automatically updating a software for the building device components.
16 . The one or more non-transitory computer-readable storage media of claim 14 , wherein training the component reliability model to produce the trained model includes determining a shape parameter and a scale parameter of a Weibull model.
17 . A predictive maintenance system, comprising:
a processing circuit including a processor and memory, the memory having instructions stored thereon that, when executed by the processor, cause the processor to:
receive warranty claim data associated with one or more building devices or building device components;
process the warranty claim data using natural language processing to generate a training data set by:
generating a plurality of clusters of warranty claims, each cluster based on a noun representing one of the building device components identified in the warranty claim as part of both one or more causes and one or more solutions associated with failure of the one or more building devices or the building device components; and
selecting, for the training set, a most frequent cause and solution of the one or more causes and the one or more solutions from a quantity of the plurality of clusters;
train a component reliability model using the training data set to produce a trained model, wherein training the component reliability model comprises configuring the component reliability model to output a reliability metric related to the one or more causes by adjusting parameters of the component reliability model; and
in response to predicting a failure of a building device or component of the one or more building devices or building device components using the component reliability model:
generate electronic control signals for communicating a command to update a software version of the building device or component; and
transmit the electronic control signals to the building device or component, causing the building device or component to initiate an update of the software version.
18 . The predictive maintenance system of claim 17 , wherein the warranty claim data includes a warranty claim comment.
19 . The predictive maintenance system of claim 18 , wherein processing the warranty claim data comprises at least one of identifying key words in the warranty claim comment, removing a stop word in the warranty claim comment, lemmatizing words in the warranty claim comment, or removing unnecessary words from the warranty claim comment.
20 . The predictive maintenance system of claim 17 , wherein the instructions further cause the processor to:
generate the reliability metric describing a predicted failure time associated with the one or more building devices or the building device components based on the trained model; and
update the building device components based on the reliability metric.