基于CNN-SVM的船舶電力推進系統故障診斷技術研究

摘 要：綜合電力推進系統是現代船舶技術的跨越式發展，對解決船舶動力平臺問題具有重要意義。為避免因電氣設備運行故障對船舶運行安全性的影響，該文研究基于卷積神經網絡和支持向量機融合的故障診斷方法。通過CNN來提取船舶電力推進系統故障信號的深層特征，將其作為故障分類器的輸入，然后由SVM分類器進行故障分類。通過仿真實驗后，發現學習率為0.001時，懲罰因子為1.5時，對應的故障診斷準確率最高，抗干擾能力較強。利用CNN和SVM融合的故障診斷方法，可有效提升船舶電力推進系統電氣設備運行的可靠性，根據船舶電氣系統運行特點，不斷完善故障診斷方法，進一步推進我國船舶技術的發展進程。

關鍵詞：船舶;電力推進系統;故障診斷;特征提取;信號

中圖分類號：U672 文獻標志碼：A 文章編號：2095-2945（2024）22-0001-04

Abstract： Integrated electric propulsion system is a great-leap-forward development of modern ship technology， which is of great significance to solve the problem of ship power platform. In order to avoid the influence of electrical equipment failure on the safety of ship operation， this paper studies the fault diagnosis method based on the fusion of convolution neural network and support vector machine. The deep features of the fault signals of the marine electric propulsion system are extracted by CNN and used as the input of the fault classifier， and then the fault is classified by the SVM classifier. Through the simulation experiment， it is found that when the learning rate is 0.001 and the penalty factor is 1.5， the corresponding fault diagnosis accuracy is the highest and the anti-interference ability is strong. The fault diagnosis method based on the integration of CNN and SVM can effectively improve the reliability of the operation of electrical equipment in marine electric propulsion system. According to the operation characteristics of marine electrical system， the fault diagnosis method can be continuously improved， and the development process of ship technology in our country can be further promoted.

Keywords： ship; electric propulsion system; fault diagnosis; feature extraction; signal

隨著推進載荷、服務載荷和任務系統等對功率需求的不斷增長，穩定可靠的船舶電力推進系統對于維持船舶的穩定運行具有重要意義。中壓直流電力推進系統作為新一代船舶電力驅動動力方式，具有變壓器尺寸小、功率密度高、高頻運行、節能降耗和電站靈活性等優勢性能，有效解決了船舶動力平臺問題。然而受到運行環境的影響，加之系統自身是一種緊密耦合和復雜的網絡結構，在船舶航行的過程中一旦電氣系統出現故障，將會威脅到船舶航行的安全性和穩定性[1]。為了提升電力推進系統運行的可靠性，應該根據船舶電力推進系統的特點，構建一套科學可靠的故障診斷方法，在船舶不同操作條件下對系統中不同類型的故障都能夠進行快速檢測和定位，為船舶電力推進系統的安全運行提供保障?！?br>