Estimation apparatus, control system, estimation method, and program
AssigneeNEC Corporation
InventorHiroshi Yoshida
Life of this patent
Estimating the term…
Abstract
In order to provide an estimation apparatus estimating a true value from observation data without determining a state equation for a target to be controlled, the estimation apparatus includes a prediction section and an update section. The prediction section executes linear prediction on time-series data, which includes observation values acquired from a sensor attached to the target to be controlled, to calculate an estimate value before update relating to a state of the target to be controlled. The update section updates the estimate value before update by using the observation value acquired from the sensor. The prediction section may calculate the estimate value before update by weighted linear prediction on the time-series data of the observation values.
Claims15
An estimation apparatus comprising:
- a memory storing instructions;
- and
- one or more processors configured to execute the instructions to, at each of a plurality of times:
- before updating an estimate value relating to a state of a target to be controlled,
- execute linear prediction on time-series data, which includes observation values acquired from a sensor attached to a target to be controlled, to calculate the estimate value;
- calculate an error covariance at a current time based on the error variance calculated after updating the estimate value at a prior time;
- calculate a Kalman gain by using the error covariance at the current time calculated before updating the estimate value;
- update the estimate value by using the update value before updating, an observation value acquired from the sensor, and the Kalman gain calculated before updating;
- after updating the estimate value relating to the state of the target to be controlled,
- recalculate the error covariance at the current time by using the error covariance at the current time calculated before updating the estimate value and the Kalman gain;
- and
- control an actuator by using the estimate value.
A system comprising:
- a sensor attached to a target to be controlled;
- an actuator;
- a memory storing instructions;
- and
- one or more processors configured to execute the instructions to, at each of a plurality of times:
- before updating an estimate value relating to a state of the,
- execute linear prediction on time-series data, which includes observation values acquired from the sensor, to calculate the estimate value;
- calculate an error covariance at a current time based on the error variance calculated after updating the estimate value at a prior time;
- calculate a Kalman gain by using the error covariance at the current time calculated before updating the estimate value;
- update the estimate value by using the update value before updating, an observation value acquired from the sensor, and the Kalman gain calculated before updating;
- after updating the estimate value relating to the state of the target to be controlled,
- recalculate the error covariance at the current time by using the error covariance at the current time calculated before updating the estimate value and the Kalman gain;
- and
- control the actuator by using the estimate value.
An method performed by a processor and comprising, each of a plurality of times:
- before updating an estimate value relating to a state of a target to be controlled,
- executing linear prediction on time-series data, which includes observation values acquired from a sensor attached to a target to be controlled, to calculate the estimate value;
- calculating an error covariance at a current time based on the error variance calculated after updating the estimate value at a prior time;
- calculating a Kalman gain by using the error covariance at the current time calculated before updating the estimate value;
- updating the estimate value by using the update value before updating, an observation value acquired from the sensor, and the Kalman gain calculated before updating;
- after updating the estimate value relating to the state of the target to be controlled,
- recalculating the error covariance at the current time by using the error covariance at the current time calculated before updating the estimate value and the Kalman gain;
- and
- controlling an actuator by using the estimate value.
Dependencies are inferred from the claim text.