What is RLS algorithm?
[Users questions | Browse:66 | Favorites]
Users questions: Original algorithm can give it? Online procedures to ~ but still want to know the original ~
Experts answer: I do not know the mathematical principle of the RLS algorithm you mean what is the English spelling, but I guess you're wondering "recursive least square algorithm" - RLS algorithm, right? :) The following is a least squares method, also known information (as opposed to Shall not easy to upload pictures, but for reference): In our study two variables (x, y) the relationship between, usually a series of pairs of data can be (x1, y1, x2, y2. .. xm, ym); to the data depicted in the xy Cartesian coordinate system (Figure 1), if Found near these points in a straight line, you can make this linear equation such as (type 1-1). Y total = a0 + a1X (type 1-1) where: a0, a1 are arbitrary real numbers for the establishment of this linear equation will determine a0 and a1, application of "least squares method", the measured value Yi and using (type 1 -1) Calculated value (Y total = a0 + a1X) deviations (Yi-Y meter) square and (Yi-Y dollars) 2 minimum "optimization criterion." Order: = (Yi-Y dollars) 2 (type 1-2) to (type 1-1) into (type 1-2) to go: = (Yi-a0-a 1Xi) 2 (type 1-3) when (Yi-Y meter) square minimum, can function on a0, a1 the partial derivative, so that the two partial derivatives equal to zero. (Eq. 1-4) (Eq. 1-5) namely: ma0 + ( Xi) a1 = Yi (type 1-6) ( Xi) a0 + ( X i2) a1 = (Xi, Yi) (type 1-7) got two on a0, a1 equations for the two unknowns, solve the two equations obtained: a0 = ( Yi) * m- a1 ( Xi) * m (type 1-8) a1 = [ XiYi-( Xi Yi) * m] * [ Xi2-( Xi) 2 * m)] (type 1-9) when the a0, a1 into (type 1-1), then the (type 1-1) line is our meta-regression equation, namely: mathematical model. In the regression process, the return is not all relational regression through each data point (x1 , Y1, x2, y2 ... xm, ym), to determine the correlation of the good or bad, can make use of the correlation coefficient "R", statistics "F", the residual standard deviation "S" to judge; "R" increasingly close to 1 as possible; "F" in the absolute value of the bigger is better; "S" more close to 0 as possible. R = [ XiYi-m ( Xi * m) ( Yi * m)] * SQR {[ Xi2-m ( Xi * m) 2] [ Yi2-m ( Yi * m) 2]} (type 1 - 10) * in (type 1-1), m is the sample size, that is, the number of experiments; Xi, Yi, respectively, a set of arbitrary Experimental X, Y values. Applications of calculus topics of a least squares method from the previous study, we know that the least square method can be used to handle a set of data can be determined from a set of data for the dependence between variables, this function is called experience formula. This project will introduce the exact least squares Definition and how to seek and linearly relationship between the empirical formula. Assumes that the measured variables between experimental data points, ..., then in the plane, we can get points, this is called a "scatter map "can be roughly seen from the figure these points scattered in a roughly straight line close by, we think that between the approximate and A linear function solving steps described below. Consider the function, where and are constants to be determined. If a line can be considered the relationship between variables. But generally speaking, these points can not be along the same line. Hutchison , which reflects the straight line to describe, when calculated with the actual values of the deviation. of course Requested deviations as small as possible, but can be positive or negative, it is not that the total deviation, the function to reflect the good relations between variables, because this time the absolute value of each deviation can be large. In order to improve the a defect to be considered for the place. However, because the absolute value is not easy for the resolution operator, therefore, further Used to measure the total deviation. Because of the square and the smallest deviation can ensure that every deviation will not be great. So the question comes down to determining the constants and, to a minimum. Use this method to determine the coefficient, the method is called least- multiplication. by the maximum principle was that the solutions of the simultaneous equations obtained (*) problems I study of a chemical Reaction process, the temperature ) on the product yield (%) of the measured data are as follows: temperature ) 100110120130140150160170180190 yield (%) 45515461667074788589 (1) the use of "ListPl
Experts answer: I do not know the mathematical principle of the RLS algorithm you mean what is the English spelling, but I guess you're wondering "recursive least square algorithm" - RLS algorithm, right? :) The following is a least squares method, also known information (as opposed to Shall not easy to upload pictures, but for reference): In our study two variables (x, y) the relationship between, usually a series of pairs of data can be (x1, y1, x2, y2. .. xm, ym); to the data depicted in the xy Cartesian coordinate system (Figure 1), if Found near these points in a straight line, you can make this linear equation such as (type 1-1). Y total = a0 + a1X (type 1-1) where: a0, a1 are arbitrary real numbers for the establishment of this linear equation will determine a0 and a1, application of "least squares method", the measured value Yi and using (type 1 -1) Calculated value (Y total = a0 + a1X) deviations (Yi-Y meter) square and (Yi-Y dollars) 2 minimum "optimization criterion." Order: = (Yi-Y dollars) 2 (type 1-2) to (type 1-1) into (type 1-2) to go: = (Yi-a0-a 1Xi) 2 (type 1-3) when (Yi-Y meter) square minimum, can function on a0, a1 the partial derivative, so that the two partial derivatives equal to zero. (Eq. 1-4) (Eq. 1-5) namely: ma0 + ( Xi) a1 = Yi (type 1-6) ( Xi) a0 + ( X i2) a1 = (Xi, Yi) (type 1-7) got two on a0, a1 equations for the two unknowns, solve the two equations obtained: a0 = ( Yi) * m- a1 ( Xi) * m (type 1-8) a1 = [ XiYi-( Xi Yi) * m] * [ Xi2-( Xi) 2 * m)] (type 1-9) when the a0, a1 into (type 1-1), then the (type 1-1) line is our meta-regression equation, namely: mathematical model. In the regression process, the return is not all relational regression through each data point (x1 , Y1, x2, y2 ... xm, ym), to determine the correlation of the good or bad, can make use of the correlation coefficient "R", statistics "F", the residual standard deviation "S" to judge; "R" increasingly close to 1 as possible; "F" in the absolute value of the bigger is better; "S" more close to 0 as possible. R = [ XiYi-m ( Xi * m) ( Yi * m)] * SQR {[ Xi2-m ( Xi * m) 2] [ Yi2-m ( Yi * m) 2]} (type 1 - 10) * in (type 1-1), m is the sample size, that is, the number of experiments; Xi, Yi, respectively, a set of arbitrary Experimental X, Y values. Applications of calculus topics of a least squares method from the previous study, we know that the least square method can be used to handle a set of data can be determined from a set of data for the dependence between variables, this function is called experience formula. This project will introduce the exact least squares Definition and how to seek and linearly relationship between the empirical formula. Assumes that the measured variables between experimental data points, ..., then in the plane, we can get points, this is called a "scatter map "can be roughly seen from the figure these points scattered in a roughly straight line close by, we think that between the approximate and A linear function solving steps described below. Consider the function, where and are constants to be determined. If a line can be considered the relationship between variables. But generally speaking, these points can not be along the same line. Hutchison , which reflects the straight line to describe, when calculated with the actual values of the deviation. of course Requested deviations as small as possible, but can be positive or negative, it is not that the total deviation, the function to reflect the good relations between variables, because this time the absolute value of each deviation can be large. In order to improve the a defect to be considered for the place. However, because the absolute value is not easy for the resolution operator, therefore, further Used to measure the total deviation. Because of the square and the smallest deviation can ensure that every deviation will not be great. So the question comes down to determining the constants and, to a minimum. Use this method to determine the coefficient, the method is called least- multiplication. by the maximum principle was that the solutions of the simultaneous equations obtained (*) problems I study of a chemical Reaction process, the temperature ) on the product yield (%) of the measured data are as follows: temperature ) 100110120130140150160170180190 yield (%) 45515461667074788589 (1) the use of "ListPl
- How to broadcast the text into a Girl?
- Shame ~ ~ I do not know
- Science Wubi
- matlab neural network using nonlinear correction ~!
- Emergency: If you are a robot with artificial intelligence to speak, you say?
- How to set up automatic camera
- On Artificial Intelligence in the use of accounting information systems!
- Consult expert Ji Jiji
- I'll give you a one doctrine. .
- Attributive participle and past participle can do it?
- This information provided by the users.Thanks!
Computer Knowledge Copyright 2005-2015 All Rights Reserved.