【正文】
node work is list in table II. Six fillins will be produced, so more memory () and more operations (321 multiply operations) are spent in the procedure of forward and backward substitution during once iteration. The total number of iterations required reduces to thirteen, which suggests that the calculation accuracy for linear equations could be raised by plete pivoting. Finally, the number of multiply operations reduces to 5573 thanks to smaller number of iterations. C. Puropse 3: Improving Accuracy while preserving the sparsity Only one small impedance branch exists in the system, so only four entries (submatrices) corresponding to node 4 and node 6 are very large in admittance matrix (Jacobin matrix). During the process of forward substitution, once node 4 or node 6 is elimination, the submatrix prised of rest elements could keep good numerical stability and numbering of rest nodes would not make a difference to the accuracy of the solution. To take both accuracy and sparsity into account, we numbered node 4 first, then numbered other nodes following the method used for purpose 1. That is what we called scheme III for the 6node work. The sequence of the node numbered for the 6node work is given in table III. 外文翻譯(原 文) 10 Since only one small impedance branch exists in the system and it connects to node 4, the degree of which is one. Scheme III will meet the request of purpose 1. So the number of fillins, memory requirements and operations needed for factorization are all the same with scheme I. Only nine iterations will be needed to insure the convergence, result in a large save of calculation (only 2107 multiply operations). The reduction on the number of iterations indicates that more exact solutions for the linear equations could be got using scheme III. After analysis and parison, the reasons are as follows: ? The diagonal element related to node 4 is just a little smaller than the one related to node 6, so eliminate node 4 first will not decrease accuracy. The scheme could meet plete pivoting approximately. ? Fewer operations in scheme III reduce the rounding error of calculator floatingpoint numbers. Especially, if eliminate node 6 first, very small value might be added to diagonal element of node 2 and node 5, which would cause serious rounding error. While, if eliminate node 4 first, a sizable value will be added to diagonal element of node 6, producing a value in the normal range. TABLE III. REORDERED NODES USING SCHEME THREE TABLE IV. PERFORMACNE OF NEWTON POWER FLOW USING DIFFERENT SCHMEMS OF NODE ORDERING 外文翻譯(原 文) 11 V. CONCLUSION Theoretical analysis and the result of numerical calculating suggest that it is necessary to consider the influence of node ordering on the accuracy of the power flow calculation. If the node ordering algorithm takes both memory and accuracy into account reasonably, the performance of power flow calculation could be further improved. Elementary conclusions of this paper are as follows: For the wellconditioning power system, the influence of node ordering on the accuracy of power flow calculation could be neglect. It is more important to focus our attention on keeping the sparsity to save memory requirement and pute operations. For the illconditioning power system, the accuracy must be considered in node ordering algorithm to speed up the convergence rate. On this basis, if the sparsity is considered meanwhile, more accuracy might be obtained because of the reduction of float point putation. VI. REFERENCES [1] Allen J. Wood and Bruce F. Wollenberg, ―Power Generation, Operation and Cotrol (Second Edition),‖ Tsinghuo University Press, 2020. [2] W. F. Tinney and J. W. Walker. ―Direct solutions of sparse work equations by optimally ordered triangular factorization,‖ Proceedings of the IEEE, vol. 55, , pp. 18011809, November 1967. [3] K. M. Sambarapu and S. M. Halpin, ―Sparse matrix techniques in power 外文翻譯(原 文) 12 systems,‖ ThirtyNinth Southeastern Symposium on System Theory, March 2020. [4] W. F. Tinney and C. E. Hart, ―Power flow solution by Newton39。在本文中,我們 試圖為更多合理性排序算法 奠定了基礎(chǔ), 這樣 可以使內(nèi)存和準(zhǔn)確性 之間進(jìn)行合理的比較 。 牛頓 — 拉夫遜 算法 是解決這個(gè)問題最常用 的方法 ,它涉及到一 系列 線性方程組重復(fù)的直接 求解 。在分解過程中,內(nèi)存中的位置可以產(chǎn)生非零 輸入,從而在原始的 雅可比矩陣 中產(chǎn)生零輸入 。雖然很難找到為確定絕對的最佳 排序的 有效的算法, 但是有著接近最好效果的一些有效算法已經(jīng)得到了實(shí)際應(yīng)用。 為了 充分利用稀疏向量方法的優(yōu)點(diǎn), 通過節(jié)點(diǎn)排序 加強(qiáng) L1 的稀疏 性是十分必要的 。許多在網(wǎng)絡(luò)問題 中的 強(qiáng)大對角線矩陣滿足上述假設(shè), 并且為了保證稀疏性的排序方法 增加 了 解決方案的準(zhǔn)確性。上面提到的所有這些事情會 使問題變得更加糟糕 。接下來 的第三部分通過 一個(gè)簡單的直流 潮流來 說明節(jié)點(diǎn)的順序可能會影響 算法的 精度 。 許多數(shù)學(xué)論文 [911]都會關(guān)注高斯消元法的完全消元法與部分消元法的區(qū)別 。在 潮流 計(jì)算 中 , 部分消元法可以再沒有任何行交匯 的情況下自動實(shí)現(xiàn) ,因?yàn)樵诖蠖鄶?shù)情況下,雅可比矩陣 的 對角占優(yōu)的功能可以保證在浮點(diǎn)運(yùn)算的數(shù)值穩(wěn)定性的。因此, 與 最大的模塊元素 有關(guān)的節(jié)點(diǎn) 往往安排在前面 以達(dá)到提高精度的目的。類似地,每一個(gè)涉及到一個(gè)節(jié)點(diǎn) 的 對角線子矩陣,往往根據(jù)他們的 行列式按照 從最小到最大 的順序進(jìn)行排列 。 當(dāng)系統(tǒng)系數(shù)變廣 時(shí) ,解的精度 幾乎不可能受 舍入誤差 的 影響 ,因此把排序?qū)τ?解決方案的準(zhǔn)確性的順序 考慮在內(nèi)是 必要的。按照原來的節(jié)點(diǎn)數(shù)量,直流 潮流 方程是: 為了 模擬計(jì)算機(jī)數(shù)值計(jì)算操作, 我們用四個(gè)有效數(shù)字來解決這個(gè)問題 。因此,牛頓 – 拉夫遜潮流方法的 性能將隨著 節(jié)點(diǎn)排序的變化而 不同。 這種方案 所需要的唯一信息是描述網(wǎng)絡(luò)節(jié)點(diǎn)分支連接模式 的 一個(gè)表。編程效率是超出了目前的工作范圍。如果一個(gè)以上的節(jié)點(diǎn)符合這個(gè)標(biāo)準(zhǔn), 選擇最原始的節(jié)點(diǎn) 。 如 果 不止一個(gè) 節(jié)點(diǎn)可以引入最少的分支 節(jié)點(diǎn) , 給那個(gè)最大節(jié)點(diǎn)度的節(jié)點(diǎn)編號 。 在求解線性方程組的過程 中 ,沒 有引進(jìn)最小填充,所以表格的 因素 和 雅可比矩陣將有完全一致的結(jié)構(gòu)。 在 每次迭代期間 前后 替代 的過程中需要 123次乘法運(yùn)算 , 整個(gè)解答過程需要 7456次乘法運(yùn)算 。為 方便起見,我們利用導(dǎo)納矩陣確定數(shù)字的順序 。因此,該 方案的 的結(jié)果可能 與 稀疏矩陣方 法和許多 引入的最小填充 下形成的節(jié)點(diǎn)編號的原則 相異 。 C 目的三:保持稀疏性的同時(shí)提高精確度 在 系統(tǒng)中只存在一個(gè)小的阻抗分支,所以相應(yīng) 于 節(jié)點(diǎn) 4 和節(jié)點(diǎn) 6 的 只有四個(gè)條目(子矩陣)是非常大的導(dǎo)納矩陣(雅可比矩陣)。 6 節(jié)點(diǎn)網(wǎng)絡(luò)的節(jié)點(diǎn)編號次序如表 3所示。為了保證收斂性,只需要 9 次迭代 ,導(dǎo)致 計(jì)算量大大減少( 僅 2107 次 乘法運(yùn)算)。 外文翻譯(譯文) 20 方案三的更 少 的 操作 次數(shù) 減少計(jì)算器浮點(diǎn)數(shù)的舍入誤差。 5 結(jié)論 理論分析和數(shù)值計(jì)算結(jié)果表明 :在潮流計(jì)算中考慮節(jié)點(diǎn)排序是十分必要的。 對于 病態(tài) 的電力系統(tǒng), 為了 加快收斂速度 ,在節(jié)點(diǎn)排序算法中必須考慮精確