Patent Grant
7136763
The present invention relates to current and voltage sensors, and more particularly methods and apparatus for improving current and voltage sensor accuracy and resolution.
Electric and hybrid vehicles include a propulsion system that typically includes an electric motor and/or an engine. Current for powering the electric motor is supplied by a battery subsystem. Key performance issues of the electric and hybrid vehicles include fuel efficiency, emissions, and drivability, which depend largely on the operation of the propulsion system.
The battery subsystem is a significant element of the propulsion system of these vehicles. An accurate state of charge (SOC) algorithm improves performance of the battery subsystem, and therefore the vehicle. The SOC algorithm requires an accurate current sensor for sensing current in the battery subsystem. The battery subsystem also requires an accurate analog-to-digital (A/D) converter that communicates with the current sensor. Other applications requiring accurate current sensing include fuel cell and supercapacitor systems.
Current sensor measurements are limited by current sensor accuracy and resolution, as well as A/D converter resolution. One conventional method for improving the accuracy of current sensor measurements uses multiple A/D converters and/or multiple current sensors. Using multiple converters and sensors increases the complexity and cost of the battery subsystem. Another conventional method uses an A/D converter having a higher resolution. However, there are limits to A/D converter resolution.
A method for estimating current or voltage in an electrical system comprises generating an analog current or voltage signal using a sensor. The signal is filtered using an analog filter. An A/D converter samples the analog signal and generates a digital signal. A matrix based on characteristics of the analog filter is formed. An estimated actual current or voltage is calculated based upon a relationship between the digital signal and the matrix.
Further areas of applicability of the present invention will become apparent from the detailed description provided hereinafter. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention.
The present invention will become more fully understood from the detailed description and the accompanying drawings, wherein:
The following description of the preferred embodiment(s) is merely exemplary in nature and is in no way intended to limit the invention, its application, or uses. For purposes of clarity, the same reference numbers will be used in the drawings to identify similar elements.
Referring now to
The current sensor 14 is implemented with current sense hardware such as hall effect sensors, sense resistors, or other appropriate hardware and/or software. An analog filter 20 receives an analog current signal 22 from the current sensor 14. In the exemplary embodiment, the analog filter 20 is a low-pass filter. However, other suitable filters may be used. An analog to digital (A/D) converter 24 converts the analog current signal 22 from the current sensor 14 into a digital current signal. The A/D converter 24 may communicate with the trigger 16 to sample the analog signal 22 in sync with inverter 10 power setting changes. Alternatively, a step inverter 26 may change power settings in a manner that causes the analog signal 22 to transition through discrete levels of the A/D converter 24.
A microprocessor 28 receives the digital current signal 30 from the A/D converter 24. The microprocessor 28 executes a current estimating algorithm to reconstruct the digital current signal 30. The current estimating algorithm results in a current estimate with a resolution that is greater than the resolution of the A/D converter 24. Additionally, the microprocessor 28 generates the trigger 16 to synchronize the sampling of the A/D converter 24 with power setting changes by the inverters 10.
Referring now to
Referring now to
Referring to
In steady-state (constant current) conditions with no noise, the accuracy of the quantized input 46 is not impaired. However, variable current and noise levels affect the accuracy of the A/D converter 24 (as shown in
Referring now to
The microprocessor 28 forms a matrix A with m rows and n columns in step 104. The matrix A is formed based upon characteristics of the analog filter 20 of
An impulse response of the analog filter 20 is h(t). For t<0, h(t) is 0. The matrix A is formed as follows:
wherein TS is the sampling rate of the quantized input 46 of the A/D converter 24 and TC is the current (most recent) A/D converter sample. In step 106, the microprocessor 28 uses the column vector Y and the matrix A to derive a vector U of actual currents. A relationship between measured currents Y, matrix A, and actual currents U is expressed as:
Y=A·U+V
wherein vector V represents quantization noise and noise due to other loads. V is not know in advance and must be estimated through statistics and probability. Noise can be described by a probability density function (a histogram of noise values over a very long period of time) and an autocorrelation function (the extent that noise values are related to previous noise values).
The microprocessor 28 calculates an estimate of an actual current estimate u using the above relationship in step 108. In the preferred embodiment, the microprocessor 28 uses a least squares method to calculate the current estimate u. The microprocessor estimates a current u based on a history of current measurements Y. The least squares method calculates the estimated current vector Û as follows:
Û=(ATA)−1AT·Y.
Matrix A is known and constant. Therefore, (ATA)−1AT can be pre-calculated by the microprocessor 28, resulting in:
Û=Ā·Y.
Additionally, this calculation may be further reduced to calculate a single current estimate u:
û=Ām·Y
wherein Ām is the last row in Ā=(ATA)−1AT. The result of this simplification is that the current estimate may be reduced to a finite impulse response (FIR) filter. The FIR filter is executed in sync with changes in inverter power settings. Although the least squares method is described, other methods are possible. Alternatively, a convex optimization function may be used to estimate the current u.
Accuracy of the current estimate may be affected by the number of power setting changes and the delay between a power setting change and the execution of the current estimate. Additionally, accuracy of the current estimating algorithm 50 may be affected by knowledge of the characteristics of the analog filter 20. However, knowledge of filter characteristics are limited by manufacturing and temperature variations. The characteristics may be derived by injecting a known signal into the analog filter 20 and estimating the necessary characteristics.
Those skilled in the art can now appreciate from the foregoing description that the broad teachings of the present invention can be implemented in a variety of forms. Therefore, while this invention has been described in connection with particular examples thereof, the true scope of the invention should not be so limited since other modifications will become apparent to the skilled practitioner upon a study of the drawings, the specification and the following claims.
| Number | Name | Date | Kind |
|---|---|---|---|
| 6344670 | Yamaguchi et al. | Feb 2002 | B1 |
| 6400158 | Van Groningen et al. | Jun 2002 | B1 |
| 6637934 | Henderson et al. | Oct 2003 | B1 |
| 20010052941 | Matsunaga et al. | Dec 2001 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20040260491 A1 | Dec 2004 | US |
