Load estimation method for electronic component

Abstract

The invention relates to a load estimation method for electronic components, in particular to daily load curve estimation for electronic components such as capacitors, resistors and relays. According to the method, on the basis of a two-step modeling method of load structure analysis, firstly, characteristic parameter estimation is carried out, including calculation of a daily average load rate and a daily minimum load rate, and then curve estimation is carried out on the basis of estimation results of the characteristic parameters. Through original data generation processing, calculation model establishment and solution, and inverse generation processing, a final estimation result is obtained. The method can improve the accuracy and reliability of load estimation, and is suitable for load management and optimization of electronic components.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic components, and in particular to a load estimation method for electronic components. Background Art

[0002] Accurately estimating the load of electronic components during use is crucial for ensuring stable equipment operation and extending component lifespan. However, existing load estimation methods, such as regression analysis, moving average, and exponential smoothing, often fail to accurately capture the complex variations in electronic component load, resulting in low accuracy and reliability. Therefore, a more effective method for estimating electronic component load is urgently needed. Summary of the Invention

[0003] The purpose of the present invention is to provide a load estimation method for electronic components to solve the problems existing in the prior art.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] Estimation of characteristic parameters;

[0006] Perform per-unit normalization on the load data of electronic components to obtain the daily load curve;

[0007] Calculate the main characteristic parameters, including daily average load rate and daily minimum load rate;

[0008] Based on load structure analysis, a multivariate linear correlation model is established to estimate future characteristic parameters.

[0009] Estimation of load curve:

[0010] Determine the benchmark curve, which can be the comprehensive analysis result of historical curves or a typical actual curve;

[0011] Generate and process the original data, including sorting and difference processing;

[0012] Establish a calculation model to minimize the difference between the estimated data and the original data;

[0013] The optimal solution is obtained through iterative solution method;

[0014] The solution result is inversely generated to obtain the final estimated result.

[0015] The beneficial effects of the present invention are:

[0016] Improved the accuracy and reliability of electronic component load estimation;

[0017] Can effectively capture the complex changes in the load of electronic components;

[0018] Applicable to load management and optimization of various electronic components such as capacitors, resistors, and relays. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is the main execution process of a load estimation method for electronic components of the present invention. DETAILED DESCRIPTION

[0020] For electronic components, including capacitors (Fenghua series aluminum electrolytic capacitors), resistors (Sart series precision resistors), relays (Yongneng series relays), etc., the estimation of daily load curve is used as an example to illustrate.

[0021] 1. Estimation of characteristic parameters

[0022] The first step of the two-step modeling approach based on load structure analysis is to estimate the characteristic parameters:

[0023] Let T = 24 (indicating the number of hours), and suppose the load data for the day to be analyzed is Pt (t = 1, 2, ..., T), where Pt is the actual load value (power in watts) at hour t. Pt is normalized by the maximum load P0 to obtain the daily load curve dt (t = 1, 2, ..., T), which yields the following relationship.

[0024] P0=max{Pt},1<=t<=T. The maximum load value of the day to be analyzed, ranging from the actual measured maximum value in Pt.

[0025] dt=Pt / P0. The normalized load curve value, ranging from (0,1].

[0026] The main characteristic parameters, daily average load rate a and daily minimum load rate b, are calculated as follows:

[0027]

[0028] b=min{dt},1<=t<=T.

[0029] Different from conventional sequence prediction methods such as regression analysis, moving average method, exponential smoothing method, etc., a prediction method based on load structure analysis is proposed here to estimate future a and b.

[0030] Let the start and end dates of the estimated sample be y1 and y2, respectively, and the start and end dates of the to-be-estimated sample be y3 and y4 (y3 = y2 + 1). Let ay and by be the daily average load rate and daily minimum load rate on day y, respectively. Assume that the load can be divided into N categories (for example, N = 4, divided into continuous load, pulse load, transient load, and standby load), let Eky be the load of the kth category on day y, and Ey be the full load, satisfying:

[0031]

[0032] For historical days, that is, when y1 <= y <= y2, Eky, ay, and by are all known; the Eky for the forecast day can be obtained through load forecasting, and the corresponding characteristic parameter forecasting method is as follows:

[0033] Normalize Eky with Ey, and we have:

[0034] Fky = Eky / Ey, k = 1, 2,..., N, y1 <= y <= y4, 1 <= k <= N; Fky represents the proportion of the kth type of load, and the value range is [0, 1], satisfying:

[0035]

[0036] The following multiple linear correlation model can be established:

[0037]

[0038] In the formula, rak and rbk are the correlation coefficients of the kth type of load with the daily average load rate and the daily minimum load rate respectively. When there are many historical days, that is, when y2 - y1 + 1 >= N, it can be solved by the least squares method, denoted as r`ak and r`bk, and thus the characteristic parameters for future days can be calculated by the following formula:

[0039]

[0040] 2. Forecasting of load curves

[0041] The second step of this method is to perform curve forecasting based on the forecasting results of characteristic parameters.

[0042] First, the reference curve needs to be determined. The curves of each historical day can be comprehensively analyzed and compared, for example, through weighted synthesis (the curves of recent days should account for a larger weight) to determine the representative curve. Or a certain actual curve with typicality can be selected as the reference curve.

[0043] Next, under the condition that the normalized values d0t (t = 1, 2,..., T) of the reference day curve and the characteristic parameters a, b (0 < b < a < 1) of the day to be forecasted are known, the load curve of this day is forecasted. Assume that the normalized value of the curve of the day to be forecasted is dt (t = 1, 2,..., T), and dt has a similar shape to d0t.

[0044] 2.1 Generation and processing of original data

[0045] To weaken the randomness of the original data and provide intermediate information for establishing a calculation model, here the original data d0t is first processed by data generation:

[0046] (1) Sorting process: sort d0t from large to small to form a sequence y0j. Correspondingly, dt is sorted and recorded as yj (j = 1, 2, ..., T). The original lower mark corresponding to the sorted two sequences is hj. Then:

[0047] 1=y01>=y02>=…>=y0T>0;

[0048] 1=y1>=y2>=…>=yT=b>0;

[0049] y0j=d0hj,j=1,2,…,T;

[0050] yj=dhj,j=1,2,…,T。

[0051] (2) Difference processing: take the difference between the two adjacent items of y0j and yj and get:

[0052] x0 i=y0 i-y0(i+1)>=0,i=1,2,…,T-1;

[0053] xi=yi-y(i+1)>=0,i=1,2,…,T-1;

[0054]

[0055] The relationship between xi and characteristic parameters is:

[0056]

[0057] b=YT.

[0058] 2.2 Computational Model

[0059] Through generative processing, the problem is transformed into minimizing the difference between xi and x0i. The calculation model is:

[0060]

[0061] satisfy:

[0062]

[0063] Xi>=0,i=1,2,…,T-1.

[0064] Let the matrix:

[0065]

[0066] Constraint Matrix

[0067] Constraint Vector

[0068] The matrix description of the problem is:

[0069] minZ=(X-X0) T (X-X0) / 2;

[0070] satisfy:

[0071] AX=b,X>=0.

[0072] 2.3 Solution of the Estimation Model

[0073] Here is a simple solution:

[0074] Introducing the Lagrange multiplier W T =[w1,w2,…,w(T-1)] and V T = [v1, v2], and let the diagonal matrix formed by wi (i = l, 2, ..., T-1) be W0 = diag{wi}, let e T =[1,1,…,1], then:

[0075] W0*e=W.

[0076] Establish the following Lagrangian function:

[0077] L(X,W0,V)=(X-X0) T (X-X0) / 2-(W0*e) T XV T (AX-b).

[0078] At the optimal point X`:

[0079] X`-X0-W0*eA T V = 0;

[0080] AX`-b=0;

[0081] W0*X`=0;

[0082] X`>=0,W0>=0.

[0083] After multiplying both ends by A on the left, we can get:

[0084] V=(AA T ) -1 [bA(X0+W0*e)].

[0085] Where, (AA T ) -1 is a constant matrix. The iterative solution process is as follows:

[0086] 1) Set the initial value W0 = 0 (zero matrix), the number of iterations q = 1, and the given convergence condition u(10 -6 );

[0087] 2) Calculate V;

[0088] 3) First calculate X`=X0+A T V, then judge each component x`i, (i = 1, 2, ..., T-1): if x`i> = 0, then set wi = 0; otherwise, set wi = -x`i, x`i = 0, thus obtaining X`, W0;

[0089] 4) Determine the following convergence condition: ||AX`-b||2 / ||b||2 <u。

[0090] Here, ||||2 represents the norm. If true, the iteration ends and the optimal solution is obtained; otherwise, set q = q + 1 and return to step 2 to continue the iteration.

[0091] The above steps only involve low-dimensional matrices and vectors, and the computational effort is very small. Generally, they can converge after only a few iterations to obtain the optimal solution X' and the corresponding Lagrange multiplier.

[0092] 2.4 Inverse generation of solution results

[0093] The following is an inverse generation process of the solution result, that is, to find dt from X` to obtain the final estimated result:

[0094] (1) Processing the deficit number generates the following recursive relationship:

[0095] y1=1;

[0096] y(i+1)=yi-x`i,i=1,2,…,T-1.

[0097] (2) Reverse sorting, that is, using the original subscript hj corresponding to the sequence to restore dt:

[0098] dhj=yj,j=1,2,…,T。

[0099] At this point, the estimation of dt by d0t,a,b is completed.

[0100] The present invention relates to a load estimation method for electronic components, particularly for estimating daily load curves for electronic components such as capacitors, resistors, and relays. This method is based on a two-step modeling approach for load structure analysis. Characteristic parameters are first estimated, including calculation of the daily average load rate and the daily minimum load rate. Curve estimation is then performed based on the estimated characteristic parameters. The final estimation result is obtained through raw data generation and processing, the establishment and solution of a calculation model, and inverse generation processing. This method can improve the accuracy and reliability of load estimation and is suitable for load management and optimization of electronic components.

[0101] The above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the scope of the present invention. The scope of protection of the present invention is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present invention within the spirit and scope of protection of the present invention, and such modifications or equivalent substitutions shall also be deemed to fall within the scope of protection of the present invention.

Claims

1. A method for estimating the load of electronic components, characterized in that: The following steps are involved: Perform per-unit normalization on the load data of electronic components to obtain the daily load curve; Calculate the main characteristic parameters, including daily average load rate and daily minimum load rate; Based on load structure analysis, a multivariate linear correlation model is established to estimate future characteristic parameters; Determine the baseline curve; Generate and process the original data, including sorting and difference processing; Establish a calculation model to minimize the difference between the estimated data and the original data; The optimal solution is obtained through iterative solution method; The solution result is inversely generated to obtain the final estimated result.

2. The method according to claim 1, characterized in that The load is divided into continuous load, pulse load, transient load and standby load.

3. The method according to claim 1, characterized in that The per-unit normalization formula is: dt=Pt / P0, where Pt is the load measured value at hour t, and P0 is the maximum load.

4. The method according to claim 1, wherein The method for establishing the multivariate linear correlation model is as follows: using Ey to per-unit-normalize Eky to obtain Fky; Eky is recorded as the load of the k-th load on day y, and Ey is the full load.

5. The method according to claim 4, characterized in that Fky=Eky / Ey,k=1,2,…,N,y1<=y<=y4; Fky represents the proportion of the kth type of load, and its value range is [0,1]. Assume that the starting and ending dates of the estimated sample are y1 and y2 respectively, and the starting and ending dates of the estimated sample are y3 and y4 respectively (y3=y2+1).

6. The method according to claim 1, characterized in that The generation process includes: sorting the original data d0t to obtain y0j, sorting the estimated data dt to obtain yj, and then performing difference processing to obtain x0i and xi.

7. The method according to claim 1, characterized in that The electronic components include capacitors, resistors, and relays.

Images