Anomaly detection and notification of electric arc furnace, system and method

A method for identifying, classifying, and sending notification of an electric arc furnace's (EAF) anomalies to improve the EAF efficiency and reliability, the method comprising the steps of:

• establishing baseline state measurements of the EAF;
• receiving new state measurements of the EAF;
• statistically testing the new state measurements against the baseline state measurements;
• identifying as an anomaly a failed statistical test;
• classifying the identified anomaly;
• sending notification of the classified anomaly to a configurable list of recipients;
• wherein the statistical test is based upon a selected one of the new state measurements;
• wherein the selected one of the new state measurements is a discrete parameter;
• parameter;
• wherein the discrete parameter is a central tendency parameter and wherein the statistical testing of the central tendency parameter includes the steps of:
• establishing the baseline measurements of the central tendency parameter over a measurement aggregation period (MAP), the MAP determined by a first heat;
• calculating a baseline central tendency of the baseline measurements of the central tendency parameter;
• calculating a dispersion parameter of the baseline measurements of the central tendency parameter;
• calculating from the new measurements of the central tendency parameter a new central tendency of the central tendency parameter for a new heat;
• and
• wherein the statistical test fails when the baseline central tendency of the new heat exceeds the baseline central tendency by a configured multiple of the dispersion parameter.

A method for identifying, classifying, and sending notification of an electric arc furnace's (EAF) anomalies to improve the EAF efficiency and reliability, the method comprising the steps of:

• establishing baseline state measurements of the EAF;
• receiving new state measurements of the EAF;
• statistically testing the new state measurements against the baseline state measurements;
• identifying as an anomaly a failed statistical test;
• classifying the identified anomaly;
• sending notification of the classified anomaly to a configurable list of recipients;
• wherein the statistical test is based upon a selected one of the new state measurements;
• wherein the selected one of the new state measurements is a discrete parameter;
• wherein the discrete parameter is a central tendency parameter and wherein the statistical testing of the central tendency parameter includes the steps of:
• calculating a central tendency of the central tendency parameter from the new measurements of the central tendency parameter over a measurement aggregation period (MAP), the MAP determined by a one of a heat, charge, number of samples and period of time;
• comparing the calculated central tendency to a constant value;
• and
• wherein the statistical test fails when the calculated central tendency exceeds the constant value.

A method for identifying, classifying, and sending notification of an electric arc furnace's (EAF) anomalies to improve the EAF efficiency and reliability, the method comprising the steps of:

• establishing baseline state measurements of the EAF;
• receiving new state measurements of the EAF;
• statistically testing the new state measurements against the baseline state measurements;
• identifying as an anomaly a failed statistical test;
• classifying the identified anomaly;
• sending notification of the classified anomaly to a configurable list of recipients;
• wherein the statistical test is based upon a selected one of the new state measurements;
• wherein the selected one of the new state measurements is a differential parameter;
• wherein the statistical testing of the differential parameter includes the steps of:
• selecting as the differential parameter to be statistically tested a differential parameter of a MAP of a one of a heat, charge, number of samples or period of time;
• establishing the baseline measurements of the discrete parameter;
• calculating a baseline central tendency of the heat to heat difference of the baseline measurements of the discrete parameter;
• calculating a baseline dispersion of the heat to heat difference of the baseline measurements of the discrete parameter;
• calculating the difference between the new measurements of the discrete parameter from a new heat and the baseline measurements of the discrete parameter;
• and
• wherein the statistical test fails when the calculated difference exceeds the baseline central tendency by a configured multiple of the calculated baseline dispersion.

A method for identifying, classifying, and sending notification of an electric arc furnace's (EAF) anomalies to improve the EAF efficiency and reliability, the method comprising the steps of:

• establishing baseline state measurements of the EAF;
• receiving new state measurements of the EAF;
• statistically testing the new state measurements against the baseline state measurements;
• identifying as an anomaly a failed statistical test;
• classifying the identified anomaly;
• sending notification of the classified anomaly to a configurable list of recipients;
• wherein the statistical test is based upon a selected one of the new state measurements wherein the selected one of the new state measurements is a differential parameter;
• wherein the statistical testing of the differential parameter includes the steps of:
• selecting as the differential parameter to be statistically tested a differential parameter of a MAP of a one of a heat, charge, number of samples or period of time;
• calculating the difference between the measurements of the discrete parameter from new measurements of the discrete parameter from the next one of the heats and the measurements of the discrete parameter of the first heat;
• and
• wherein the statistical test fails when the calculated difference exceeds a constant value.

A system for continuous monitoring of an Electric Arc Furnace (EAF) to identify and classify EAF anomalies to improve the EAF efficiency, the system comprising:

• a device configured to:
• receive and establish baseline state measurements of the EAF;
• receive new state measurements of the EAF;
• perform statistical testing of the baseline state measurements against the new state measurements;
• wherein the new state measurements may be one of a continuous parameter, a discrete parameter, and a differential parameter;
• wherein where the new state measurements is the continuous parameter the statistical testing of the continuous parameter includes the steps of:
• dividing the baseline state measurements and new state measurements of the continuous parameter into baseline bins and new measurement bins according to a time base;
• comparing each of the new measurement bin statistics of the continuous parameter to the baseline bin statistics of the continuous parameter;
• and
• wherein the statistical test fails where the new measurement bin statistics of the continuous parameter exceeds predefined statistical limits of the baseline bin statistics;
• or
• wherein where the selected one of the new state measurements is a discrete parameter, the discrete parameter is a central tendency parameter, the statistical testing of the central tendency parameter includes the steps of:
• establishing the baseline measurements of the central tendency parameter over a measurement aggregation period (MAP), the MAP determined by a first heat;
• calculating a baseline central tendency of the baseline measurements of the central tendency parameter;
• calculating a dispersion parameter of the baseline measurements of the central tendency parameter;
• calculating from the new measurements of the central tendency parameter a new central tendency of the central tendency parameter for a new heat;
• and
• wherein the statistical test fails when the baseline central tendency of the new heat exceeds the baseline central tendency by a configured multiple of the dispersion parameter;
• or
• wherein where the selected one of the new state measurements is a differential parameter, the statistical testing of the differential parameter includes the steps of:
• selecting as the differential parameter to be statistically tested a differential parameter of a MAP of a one of a heat, charge, number of samples or period of time,
• establishing the baseline measurements of the discrete parameter;
• calculating a baseline central tendency of the heat to heat difference of the baseline measurements of the discrete parameter;
• calculating a baseline dispersion of the heat to heat difference of the baseline measurements of the discrete parameter;
• calculating the difference between the new measurements of the discrete parameter from a new heat and the baseline measurements of the discrete parameter;
• and
• wherein the statistical test fails when the calculated difference exceeds the baseline central tendency by a configured multiple of the calculated baseline dispersion;
• identify as an anomaly a failed statistical test;
• classify the identified anomaly;
• and
• notify a configurable list of recipients of the classification.

An electric arc furnace (EAF), the EAF comprising:

• a refractory-lined vessel covered with a retractable roof, through which one or more graphite electrodes enter the EAF;
• a power system, powering the EAF;
• at least a first electrical metering device communicably coupled to the power system;
• an EAF process and control system measuring and controlling the EAF;
• at least a first programmable logic controllers (PLC) communicably coupled to the EAF process and control system;
• a computing device;
• the computing device communicably coupled to the first electrical metering device and the PLC;
• wherein the computing device is configured to:
• receive and establish baseline state measurements of the EAF;
• receive new state measurements of the EAF;
• perform statistical testing of the baseline state measurements against the new state measurements;
• wherein the new state measurements may be one of a continuous parameter, a discrete parameter, and a differential parameter;
• wherein where the new state measurements is the continuous parameter the statistical testing of the continuous parameter includes the steps of:
• dividing the baseline state measurements and new state measurements of the continuous parameter into baseline bins and new measurement bins according to a time base;
• comparing each of the new measurement bin statistics of the continuous parameter to the baseline bin statistics of the continuous parameter;
• and
• wherein the statistical test fails where the new measurement bin statistics of the continuous parameter exceeds predefined statistical limits of the baseline bin statistics;
• or
• wherein where the selected one of the new state measurements is a discrete parameter, the discrete parameter is a central tendency parameter, the statistical testing of the central tendency parameter includes the steps of:
• establishing the baseline measurements of the central tendency parameter over a measurement aggregation period (MAP), the MAP determined by a first heat;
• calculating a baseline central tendency of the baseline measurements of the central tendency parameter;
• calculating a dispersion parameter of the baseline measurements of the central tendency parameter;
• calculating from the new measurements of the central tendency parameter a new central tendency of the central tendency parameter for a new heat;
• and
• wherein the statistical test fails when the baseline central tendency of the new heat exceeds the baseline central tendency by a configured multiple of the dispersion parameter;
• or
• wherein where the selected one of the new state measurements is a differential parameter, the statistical testing of the differential parameter includes the steps of:
• selecting as the differential parameter to be statistically tested a differential parameter of a MAP of a one of a heat, charge, number of samples or period of time,
• establishing the baseline measurements of the discrete parameter;
• calculating a baseline central tendency of the heat to heat difference of the baseline measurements of the discrete parameter;
• calculating a baseline dispersion of the heat to heat difference of the baseline measurements of the discrete parameter;
• calculating the difference between the new measurements of the discrete parameter from a new heat and the baseline measurements of the discrete parameter;
• and
• wherein the statistical test fails when the calculated difference exceeds the baseline central tendency by a configured multiple of the calculated baseline dispersion;
• identify as an anomaly a failed statistical test;
• classify the identified anomaly;
• and
• send notification of the classified anomaly to a configurable list of recipients.