混合量子基因演算法與類神經網路作短期負載預測

Hybrid Quantum Genetic Algorithm and Artificial Neural Network for Short Term Load Forecasting

邱士齊(Shih-Chi Chiu)

指導教授 ： 曹大鵬

國立臺北科技大學/電資學院/電機工程系研究所/碩士(2012年)

https://doi.org/10.6841/NTUT.2012.00118

摘要

電力負載預測是電力公司規劃未來之電力供給時，一項不可或缺的考量因素，準確的負載預測除了可以提供適當的發電排程和規劃、降低運轉成本、維持供電可靠度，並能避免限電危機或資源浪費，進一步增加電力系統供電品質和電力公司營運的競爭力。
　　本論文以量子基因演算法(Quantum Genetic Algorithm, QGA)結合類神經網路(Artificial Neural Network, ANN)中的倒傳遞網路法(Back-Propagation Neural Network, BPNN)來求解負載預測之問題，本論文乃以倒傳遞網路為主體，然後使用量子基因演算法來求取倒傳遞網路之各項參數值，以改善原本用於求解網路最佳參數值之倒傳遞法方式。
　　在求解當中，量子基因演算法是一種新興的優化技術，該算法採用量子機率向量的編碼方式，同時使用量子位元、量子疊加態的思想，其中量子疊加態的特性能使排列更多元化，而機率表達的特性，是將解的狀態以一定的機率表達出來，能有效提高整體最佳解的搜索能力。
　　在本論文中以台電提供之電力負載資料為基準，並使用中央氣象局所提供的氣象資訊作實際之負載預測，所測得的結果並與傳統方法做比較，由結果顯示本方法可求得較小之負載預測誤差值，故本方法非常適用於實際之負載預測工作上。

關鍵字

量子基因演算法 ； 類神經網路 ； 倒傳遞類神經網路 ； 負載預測

並列摘要

Power load forecasting is one of the essential factors of electric utilities when planning the future electricity supplies. Precise forecasting of electricity consumption may not only provide proper generation commitment and scheduling but also reduce the operational cost. It can also maintain the reliability of power systems, avoiding electricity crisis or wasting resources. It can further increase the power quality of power systems and the competitiveness of utilities.
This thesis presents a hybrid Quantum Genetic Algorithm (QGA) and Back-Propagation Neural Network (BPNN) of Artificial Neural Network (ANN) for load forecasting solution. A Back-Propagation Neural Network is used for the initial load forecasting, then we used QGA approach to find the optimal solution of the parameters of BPNN to improve the existing method of Back-Propagation.
The Quantum Genetic Algorithm is a new optimization technique which uses the coding method of quantum probability vector, and also uses the quantum bit and quantum superposition at the same time. The superposition can let it express more states. The probability expression characteristic can be expressed the solution state by certain probability. It can raise the ability of optimal solution.
In this thesis we apply the power loads data from Taipower Company and the weather information from Central Weather Bureau to forecast the loads. We used the QGA-BPNN to examine if we could improve the tradition methods. The results demonstrated that QGA-BPNN is more accuracy and efficiency; thus it can be applied to the actual load forecasting.

並列關鍵字

Quantum Genetic Algorithm ； Artificial Neural Network ； Back-Propagation Neural Network ； Load Forecasting

參考文獻

[22] 施順鐘，應用類神經網路於醫院空調短期電力預測，碩士論文，國立台北科技大學電機工程研究所，台北市，2005。

[28] 王鵬翔，應用量子基因演算法求解火力機組排程，碩士論文，國立台北科技大學電機工程研究所，台北市，2011。

[8] H. T. Yang, T. C. Liang, K. R. Shih and C. L. Huang, “Power System Yearly Peak Load Forecasting: A Grey System Modeling Approach,” IEEE Proc. Energy Management and Power Delivery, Vol. 1, 1995, pp. 261-266.

[3] A. D. Papalexopulos and T. C. Hesterberg, “A Regression-Based Approach to Short-term System Load Forecasting,” IEEE Transactions on Power Systems, Vol. 5, No. 4, November 1990, pp. 1535-1547.

[4] T. Haida and S. Muto, “Regression Based Peak Load Forecasting Using a Transformation Technique,” IEEE Transactions on Power Systems, Vol. 9, No. 4, November 1994, pp. 1788-1794.