劉行 朱燕飛 于家浩
摘? 要: 針對在統(tǒng)計(jì)特性未知欺騙攻擊下,線性時變參數(shù)網(wǎng)絡(luò)化控制系統(tǒng)(NCS)容易出現(xiàn)跟蹤偏差的問題,提出了一種基于集員估計(jì)的控制策略,實(shí)現(xiàn)了系統(tǒng)狀態(tài)對參考狀態(tài)的有效跟蹤.與卡爾曼估計(jì)等方法相比,無需提前知道攻擊的統(tǒng)計(jì)特性,利用一種新型的集員估計(jì)方法,將傳感器到控制器通道和控制器到執(zhí)行器通道遭受的欺騙攻擊建模為未知但有界(UBB)的信號,其中系統(tǒng)真實(shí)狀態(tài)和估計(jì)狀態(tài)都存在于估計(jì)集中.該方法提高了估計(jì)的準(zhǔn)確性,并擴(kuò)大了適用范圍.采用Lyapunov函數(shù),給出了誤差閉環(huán)系統(tǒng)穩(wěn)定的充分條件和基于線性矩陣不等式的估計(jì)器和控制器的求解算法,并使用凸優(yōu)化方法獲得最佳的估計(jì)集.用數(shù)值仿真驗(yàn)證所提出方法的可行性.
關(guān)鍵詞: 網(wǎng)絡(luò)化控制系統(tǒng)(NCS); 集員估計(jì); 跟蹤控制; 欺騙攻擊
中圖分類號: TP 273; TN 911.7? 文獻(xiàn)標(biāo)志碼: A? 文章編號: 1000-5137(2022)02-0157-14
LIU Hang, ZHU YanfeiYU Jiahao
(College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 201418, China)
Considering the tracking deviation of the linear variable parameter networked control system(NCS) under a deception attack with unknown statistical characteristics, a novel control strategy based on set-membership estimation was proposed to realize the effective tracking of the system state to the reference system state. Firstly, compared to methods such as the Kalman estimation with the known statistical characteristics of the attack in advance, a novel set-membership estimation was proposed to improve the accuracy and application scope of the estimation, in which the deception attack suffered in the sensor-to-controller channel and the controller-to-actuator channel only needed to be modeled as an unknown but bounded(UBB) signal, and to guarantee that the true state and estimated state of the system consisted in the estimated set. Secondly, the sufficient conditions for the stability of the error closed-loop system based on the Lyapunov function were given. Simultaneously, the solving algorithm of the estimator and the controller based on the linear matrix inequality was obtained, and the convex optimization method was adopted to achieve the best estimation set. Finally, the feasibility of the proposed method was verified by numerical simulation.
networked control system(NCS); set-membership estimation; tracking control; deception attacks
0? 引 言
隨著互聯(lián)網(wǎng)以及無線通信的飛速發(fā)展,物理設(shè)備之間的通信距離不再局限于本地,先進(jìn)的通信網(wǎng)絡(luò)可以為分布于不同地點(diǎn)的兩臺或多臺物理設(shè)備,提供快速可靠的通信方案.在此背景下,網(wǎng)絡(luò)化控制系統(tǒng)(NCS)應(yīng)運(yùn)而生,并在近幾年廣泛應(yīng)用于交通、電力以及機(jī)器人遠(yuǎn)程控制系統(tǒng)等領(lǐng)域.跟蹤控制作為NCS的一個研究方向,受到越來越多的人關(guān)注,其主要目標(biāo)是保證被控對象的狀態(tài)可以較好地跟蹤參考狀態(tài),在無人機(jī)及機(jī)器人軌跡跟蹤控制等方面應(yīng)用廣泛.但是基于公共網(wǎng)絡(luò)的通信方式也給跟蹤控制帶了一些挑戰(zhàn),例如,網(wǎng)絡(luò)帶寬的有限性和不可避免的網(wǎng)絡(luò)擁塞通常會引起一些網(wǎng)絡(luò)誘導(dǎo)時延、數(shù)據(jù)丟包等問題,進(jìn)而影響NCS的跟蹤效果.除此之外,一些人為的惡意攻擊也會破壞NCS傳輸數(shù)據(jù)的完整性和準(zhǔn)確性,一定程度上來說,刻意的網(wǎng)絡(luò)攻擊是一種更為嚴(yán)重的破壞.因此,采取有效的措施,避免惡意的網(wǎng)絡(luò)攻擊以及保證NCS在被攻擊后的快速恢復(fù)能力至關(guān)重要.
目前的大多數(shù)研究把常見的網(wǎng)絡(luò)攻擊分為3大類:拒絕服務(wù)(DoS)攻擊、重放(reply)攻擊及欺騙(deception)攻擊.對于DoS攻擊,攻擊者會阻塞通信信道,最終導(dǎo)致帶寬資源被耗盡;對于reply攻擊,攻擊者會不斷重發(fā)之前已經(jīng)發(fā)送過的數(shù)據(jù),導(dǎo)致控制器不能實(shí)時接收到當(dāng)前時刻的消息;欺騙攻擊和前兩種攻擊不同,攻擊者會在傳輸?shù)恼鎸?shí)數(shù)據(jù)中注入虛假數(shù)據(jù),偽裝成真實(shí)數(shù)據(jù),因此欺騙攻擊是一種更隱蔽、破壞性更強(qiáng)的攻擊.目前,只有少部分文獻(xiàn)考慮了網(wǎng)絡(luò)攻擊下的跟蹤控制問題,如TANG 等考慮存在DoS攻擊的情況下,制定一種事件觸發(fā)條件,保證移動機(jī)器人的跟蹤收斂以及網(wǎng)絡(luò)資源的有效利用;JIANG等研究了一類存在故障和DoS攻擊下,不確定系統(tǒng)的分散自適應(yīng)模糊跟蹤控制問題;FENG 等研究了基于隨機(jī)馬爾可夫過程的多智能體系統(tǒng)的分布式安全一致性跟蹤控制問題,通過建立一種混合隨機(jī)安全控制框架,實(shí)現(xiàn)了均方指數(shù)一致性跟蹤;WAN等在異步攻擊場景下,設(shè)計(jì)了一種選擇反饋增益矩陣和耦合強(qiáng)度的算法,以確保能夠?qū)崿F(xiàn)異構(gòu)網(wǎng)絡(luò)系統(tǒng)的一致性跟蹤.目前研究主要集中使用伯努利隨機(jī)過程等模型,描述攻擊或使用卡爾曼濾波進(jìn)行狀態(tài)估計(jì),以上方法都需要提前假設(shè)攻擊的先驗(yàn)知識,如概率分布、期望等,而在實(shí)際問題中,攻擊的先驗(yàn)知識很難準(zhǔn)確被獲取.因此,建立合適的網(wǎng)絡(luò)攻擊模型是保證實(shí)現(xiàn)準(zhǔn)確跟蹤的重要基礎(chǔ).未知但有界(UBB)信號模型是一種只要求信號有界,并對信號統(tǒng)計(jì)特性沒有限制的模型,經(jīng)常用來描述過程和測量噪聲.事實(shí)上,一些欺騙攻擊也是能量有界的,它不會持續(xù)攻擊傳輸信號,每次攻擊幅度也不會無限大,因此本文作者使用UBB信號描述欺騙攻擊,與需要預(yù)知攻擊先驗(yàn)知識的模型相比,所研究的模型具有更加廣泛的適用范圍.
另外,UBB信號經(jīng)常在集員估計(jì)中被用于提高集員估計(jì)的估計(jì)精度.集員估計(jì)是近幾年被大家廣泛討論的熱點(diǎn)話題,其估計(jì)結(jié)果不是單個估計(jì)點(diǎn),而是包含系統(tǒng)真實(shí)狀態(tài)的估計(jì)集合,集合中的任意值都可以作為在合理范圍內(nèi)系統(tǒng)真實(shí)狀態(tài)的估計(jì)值,與一些點(diǎn)估計(jì)方法相比(例如卡爾曼估計(jì)),這種靈活的估計(jì)方法大大降低了對傳感器測量精度的要求.集員估計(jì)作為一種切實(shí)可行的估計(jì)方法,已經(jīng)應(yīng)用在各個領(lǐng)域中,ZHOU等使用擴(kuò)展非線性集員濾波器對多傳感器數(shù)據(jù)進(jìn)行融合,實(shí)現(xiàn)室內(nèi)移動機(jī)器人的長距離實(shí)時定位;MOUSAVINEJAD等通過計(jì)算集員估計(jì)集是否存在交并集,判斷NCS中是否存在網(wǎng)絡(luò)攻擊;TANG等提出了一種基于未知輸入的集員估計(jì)器,對執(zhí)行器故障進(jìn)行診斷.目前把集員估計(jì)使用在NCS跟蹤控制問題上的研究較少,MOUSAVINEJAD 等提出了一種彈性集員估計(jì)策略,保護(hù)系統(tǒng)不受網(wǎng)絡(luò)攻擊,并保證系統(tǒng)狀態(tài)和參考狀態(tài)都包含在估計(jì)集合內(nèi),從而實(shí)現(xiàn)狀態(tài)跟蹤.除此之外,在跟蹤控制問題上,很少有人利用集員估計(jì)集為跟蹤控制提供可靠的估計(jì)狀態(tài).在實(shí)際應(yīng)用中,當(dāng)估計(jì)的狀態(tài)從估計(jì)器傳到控制器時,往往會出現(xiàn)傳輸錯誤,如果使用估計(jì)集合中特定點(diǎn)作為控制率,將很難保證估計(jì)點(diǎn)的有效性.
基于以上研究現(xiàn)狀,本文作者使用UBB信號模型、集員估計(jì)理論和Lyapunov穩(wěn)定性理論,研究欺騙攻擊下NCS的跟蹤控制問題,找到保證集員估計(jì)有效性的充分條件.在此基礎(chǔ)上,建立誤差閉環(huán)系統(tǒng),分析閉環(huán)系統(tǒng)的穩(wěn)定性,從而確保所考慮的系統(tǒng)狀態(tài)對參考狀態(tài)的有效跟蹤.主要工作如下:
設(shè)計(jì)了一種基于集員估計(jì)的跟蹤控制策略,估計(jì)狀態(tài)使用集員估計(jì)集中的任意值而不是特定的估計(jì)點(diǎn),同時系統(tǒng)的真實(shí)狀態(tài)和任意估計(jì)狀態(tài)都包含在集員估計(jì)集中,保證了估計(jì)點(diǎn)的有效性.
建立誤差閉環(huán)系統(tǒng),把在受到UBB信號欺騙攻擊下,系統(tǒng)的狀態(tài)跟蹤問題轉(zhuǎn)化為閉環(huán)系統(tǒng)的有限時間—集員估計(jì)輸入到狀態(tài)穩(wěn)定性(FT-smISS)問題,保證系統(tǒng)跟蹤的有效性.
1? 問題描述及初步準(zhǔn)備
本研究的目標(biāo)是針對NCS設(shè)計(jì)一個基于集員估計(jì)的跟蹤控制策略,當(dāng)傳感器到控制器通道和控制器到執(zhí)行器通道存在欺騙攻擊時,所考慮的系統(tǒng)狀態(tài)可以有效地實(shí)現(xiàn)對參考狀態(tài)的跟蹤.NCS的結(jié)構(gòu)框圖如圖1所示.
系統(tǒng)描述
基于集員估計(jì)的跟蹤控制協(xié)議
集員估計(jì)跟蹤控制問題
4? 結(jié) 論
考慮了在UBB欺騙攻擊下線性時變參數(shù)的NCS,設(shè)計(jì)了一種基于集員估計(jì)的控制策略,使用估計(jì)集中的任意值代替某一特定值作為跟蹤控制策略中的估計(jì)狀態(tài),提高了估計(jì)的有效性.構(gòu)造了一個誤差閉環(huán)系統(tǒng),將被控系統(tǒng)的狀態(tài)跟蹤問題轉(zhuǎn)化為誤差閉環(huán)系統(tǒng)的穩(wěn)定性問題,提出了保證集員估計(jì)可用性和狀態(tài)跟蹤有效性的充分條件.最后討論了3種情況下系統(tǒng)的跟蹤效果,驗(yàn)證了所提方法的可行性.
參考文獻(xiàn):
[1]? LI S B, WANG Z, ZHANG W D, et al. Status and prospect of networked control system [J]. Information and Control, 2003,32(3):239-244.
[2]? BAI T, WU Z M, YANG G K. Networked control systems(NCSs) [J]. Control Theory & Applications,2004,21(4):584-590.
[3]? JIANG L B, HONG S. Research on the application of intelligent vehicle path tracking control [J]. Machine Design and Manufacturing Engineering,2021,50(6):51-55.
[4]? ZHANG S J, LI S, YAN P, et al. Switching LPV robust tracking control for helicopters [J]. Journal of Nanjing University of Aeronautics & Astronautics,2021,53(2):260-266.
[5]?; WANG K Y, WEI L N, TIAN E G, et al. Memory?event?triggered control of networked control systems subject to DoS attacks [J]. Information and Control,2019,48(5):528-535.
[6]? HUANG L, GUO J, ZHANG H Y. Observer?based dynamic event triggering control for networked systems with periodic denial?of?service attack [J]. Control Theory & Applications,2021,38(6):851-861.
[7]? FRANZ G, TEDESCO F, FAMULARO D. Resilience against replay attacks: a distributed model predictive control scheme for networked multi?agent systems [J]. IEEE/CAA Journal of Automatica Sinica,2021,8(3):628-640.
[8]? YANG J J, ZHANG Z D, XIE L B. Coding detection scheme for replay attack based on the control performance index [J]. Information and Control,2021,50(3):329-336.
[9]? DING D R, WANG Z D, HO D W C,et al. Distributed recursive filtering for stochastic systems under uniform quantizations and deception attacks through sensor networks [J]. Automatica,2017,78:231-240.
[10] LI F Q, GAO L S, ZHENG B Z, et al. Event?triggered secure control for networked systems under deception attacks [J]. Computer Engineering and Applications,2021,57(5):264-270.
[11] TANG Y, ZHANG D, HO D W C, et al. Event?based tracking control of mobile robot with denial?of?service attacks [J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems,2018,50(9):3300-3310.
[12] JIANG X, MU X, HU Z. Decentralized adaptive fuzzy tracking control for a class of nonlinear uncertain interconnected systems with multiple faults and denial?of?service attack [J]. IEEE Transactions on Fuzzy Systems,2021,29(10):3130-3141.
[13] FENG Z, WEN G, HU G. Distributed secure coordinated control for multiagent systems under strategic attacks [J]. IEEE Transactions on Cybernetics,2017,47(5):1273-1284.
[14] WAN Y, CAO J D. Observer?based tracking control for heterogeneous dynamical systems under asynchronous attacks[C]// 2017 International Workshop on Complex Systems and Networks (WCSN). Doha: IEEE,2017:224-229.
[15] DING D R, SHEN Y X, SONG Y, et al. Recursive state estimation for discrete time?varying stochastic nonlinear systems with randomly occurring deception attacks [J]. International Journal of Genaral Systems,2016,45(5):548-560.
[16] WEI G L, LIU S, WANG L C, et al. Event?based distributed set?membership ?ltering for a class of timevarying non-linear systems over sensor networks with saturation effects [J]. International Journal of General Systems,2016,45(5):532-547.
[17] WENG P D, CHEN B, YU L. Fusion estimate of FDI attack signals [J]. Acta Automatica Sinica,2021,47(9):2292-2300.
[18] ZHANG Y L, ZHU Y F, FAN Q Q. A novel set?membership estimation approach for preserving security in networked control systems under deception attacks [J]. Neurocomputing,2020,400(4):440-449.
[19] ZHOU B, QIAN K, MA X D, et al. Multi?sensor fusion for mobile robot indoor localization based on a set?membership estimator [J]. Control Theory & Applications,2017,34(4):541-550.
[20] MOUSAVINEJAD E, YANG F, HAN Q L, et al. A novel cyber attack detection method in networked control systems [J]. IEEE Transactions on Cybernetics,2018,48(11):3254-3264.
[21] TANG W T, WANG Z H, WANG Y, et al. Fault diagnosis for uncertain systems based on unknown input set-membership filters [J]. Acta Automatica Sinica,2018,44(9):1717-1724.
[22] MOUSAVINEJAD E, GE X, HAN Q L, et al. Resilient tracking control of networked control systems under cyber attacks[J]. IEEE Transactions on Cybernetics,2021,51(4):2107-2119.
(責(zé)任編輯:包震宇,馮珍珍)