Multi-target fault-disturbance cooperative diagnosis method for intelligent traffic signal system

Abstract

The invention discloses a multi-target fault-disturbance collaborative diagnosis method for an intelligent traffic signal system. According to the method, firstly, a switching positive system state space model is established for an intelligent traffic signal system, then a fault diagnosis filter and an augmented dynamic system based on a dynamic event triggering mechanism are designed, and modal dependence minimum dwell time MDMDT constraint is introduced to relieve the problem of communication resource waste. And furthermore, a linear programming method is utilized to optimize the stable operation condition of the augmented dynamic system, so that the designed fault diagnosis filter can provide positive asymptotic estimation, and not only has robustness to external disturbance, but also has relatively high sensitivity to faults. The multi-target fault-disturbance collaborative diagnosis method provided by the invention can effectively improve the utilization rate of system resources, optimize the system performance, realize stable and efficient diagnosis, improve the safety of urban traffic, and has very important significance for the control theory and practical application.

Description

Technical Field

[0001] The present invention belongs to the fields of automation technology and modern control, and relates to the fault diagnosis of urban intelligent transportation systems. Specifically, it relates to a multi-objective fault-disturbance collaborative diagnosis method for intelligent traffic signal systems. Background Art

[0002] As a core component of modern urban traffic management, the stable operation of intelligent traffic signal systems is crucial for ensuring the quality of residents' lives and driving safety. However, with the acceleration of urbanization and the continuous increase in the number of motor vehicles, urban traffic flow has become increasingly complex, and the reliability and complexity of intelligent traffic signal systems are facing severe challenges. To cope with the complex traffic environment, intelligent traffic signal systems integrate a large number of sensors, communication devices, and complex algorithm modules. During long-term operation, these system components will inevitably experience various faults and disturbances. In addition, faults and disturbances in the system often have coupling and concurrency, that is, multiple faults or disturbances occur simultaneously and there are complex interactions between them. These problems not only affect the real-time control performance of intelligent traffic signal systems but may also lead to traffic flow disorders and even serious safety hazards. Therefore, it is crucial to detect and resolve these faults and disturbances in a timely manner to ensure the normal operation of intelligent traffic signal systems.

[0003] In the case of limited communication resources, the time-triggered mechanism updates data at fixed time intervals. Even if the system state does not change significantly, it will continue to transmit data, increasing unnecessary communication burdens. Although the static event-triggered mechanism reduces redundant transmissions to a certain extent, in the face of a complex and dynamic traffic environment, it may still lead to overly conservative trigger conditions, affecting the efficiency of fault diagnosis and also resulting in waste of communication resources. Also, due to physical limitations, traffic lights must stay on for a certain period after being activated, and the switching rules need to have more stringent time limits to ensure the stability of traffic order. The average dwell time (ADT) constraint is difficult to accurately characterize the dwell characteristics of traffic lights, and this mismatch may affect the correct assessment of the system state and reduce the accuracy of fault diagnosis.

[0004] In addition, in the urban traffic signal fault diagnosis system, due to the non-negative traffic flow constraint of the traffic signal system and the different red-green light patterns at intersections, if the model cannot accurately describe the dynamic characteristics of the entire system, it may lead to deviations in the diagnosis results. Therefore, it is urgent to further optimize the fault diagnosis strategy of urban traffic signal systems to improve the diagnosis accuracy and the stability of system operation. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technologies, the present invention proposes a multi-objective fault-disturbance collaborative diagnosis method for an intelligent traffic signal system. A switched positive system is selected to model the intelligent traffic signal system, and a switching signal based on the mode-dependent minimum dwell time (MDMDT) constraint is proposed. By designing an L - / L1 fault diagnosis filter, the multi-objective fault-disturbance collaborative diagnosis of the intelligent traffic signal system is realized, and problems such as waste of network resources and potential traffic safety hazards are solved.

[0006] A multi-objective fault-disturbance collaborative diagnosis method for an intelligent traffic signal system comprises the following specific steps:

[0007] Step 1. Establish a switched positive system state-space model of the intelligent traffic signal system

[0008]

[0009] where x(t) ∈ R n represents the system state, n represents the number of buffer zones at the intersection, represents the first derivative of x(t), y(t) ∈ R p represents the measurement output, p represents the number of flow observation functions, w(t) ∈ R w represents the disturbance input caused by sensor errors and lane-changing at adjacent lanes, etc., w represents the number of disturbances, f(t) ∈ R f represents the fault input in the intersection buffer zone, f represents the number of faults. σ(t): [t0, ∞) → N ={1, 2, …, N} is the switching signal of the switched system, N represents the number of subsystems, σ(t) = i ∈ N represents that the i-th subsystem is activated, where N represents a finite set of the number of subsystems. x0 and t0 represent the initial state and the initial time of the system respectively. A i ∈ R n×n 、D i ∈ R n×w 、F i ∈ R n×f 、C i ∈ R p×n are known system matrices determined by the intersection structure.

[0010] Step 2. Construct a fault diagnosis filter for the system under a dynamic event-triggering mechanism and MDMDT constraint

[0011] s2.1. Design a fault diagnosis filter for the system under a dynamic event-triggering mechanism:

[0012]

[0013] Among them, represents the estimated vector of x(t), and r(t) ∈ R l represents the residual signal, and l represents the number of residual signals. is the unknown filter parameter matrix. is the sampled signal at the h-th dynamic event triggering moment of the system, h ∈ Z, and Z represents the set of natural numbers. Based on this, the following dynamic event triggering mechanism is established:

[0014]

[0015] Among them, η(t) ≥ 0 represents the internal dynamic variable, θ > 0 represents the participation degree of η(t). As θ increases, the influence of η(t) on the system gradually decreases. ξ > 0 represents the attenuation rate of the filter. As ξ increases, the attenuation rate increases, meaning that fewer signal components are retained after filtering. 0 ≤ ψ i < 1 represents the tolerance of the dynamic event triggering mechanism. As ψ i increases, the tolerance for the error gradually increases, meaning that the amount of data transmitted will decrease. Ω i ∈ R p represents the unknown dynamic event triggering vector. The superscript T represents the transpose. inf{} represents the infimum.

[0016] Considering the positivity constraint of the system, the sampling error is redefined as:

[0017]

[0018] By choosing an appropriate ε such that e y (t) ≥ 0, the updated dynamic event triggering mechanism is as follows:

[0019]

[0020] Under this dynamic event triggering mechanism, define According to the switched positive system state space model of the intelligent transportation signal system, the following augmented switched positive system is obtained:

[0021]

[0022] Among them, 0 n×n 、0 n×p 、0 n×w 、0 n×f represent the n×n, n×p, n×w, and n×f matrices with all elements being 0, respectively.

[0023] s2.2. Design the minimum dwell time constraint of modal dependence MDMDT:

[0024]

[0025] Among them, T ki represents the dwell time when the i-th subsystem is activated in the switching interval [t k , t k+1 ), and represents the minimum dwell time of modal dependence when the i-th subsystem is activated.

[0026] Based on the MDMDT constraint, the switching interval [t k , t k+1 ) is divided into and two parts, and the interval is divided into S parts, and the length of each part is The s-th interval is expressed as

[0027] Select the following discrete linear copositive Lyapunov function:

[0028]

[0029] Among them is a time-dependent piecewise linear function, represents that the function values are positive real numbers and the number is 2n:

[0030]

[0031] Among them, represents the value of the piecewise function v i (t) at the critical point. Take the derivative of v i (t):

[0032]

[0033] s2.3. Based on the dynamic event-triggering mechanism in s2.1 and the MDMDT constraint in s2.2, modify the fault diagnosis filter to:

[0034]

[0035] Furthermore, a new augmented system can be obtained:

[0036]

[0037] Among them, is modal-dependent and piecewise, and It is also mode-dependent and segmented.

[0038] Step 3, solve the fault diagnosis filter gain matrix:

[0039] Design the conditions for the stable operation of the augmented dynamic system, analyze the positivity, stability and L - / L1 performance conditions, and design the parameter matrix of the fault diagnosis filter as:

[0040]

[0041] Among them, the constant φ i > 0, the vector I n represents the n×n identity matrix, 1 n represents the n-dimensional column vector with all elements being 1, represents the n-dimensional column vector with the ι-th row element being 1 and the remaining elements being 0, that is The superscript represents the pseudo-inverse of the matrix.

[0042] Step 4, multi-objective fault-disturbance co-diagnosis of the intelligent transportation signal system

[0043] Design the residual evaluation function J r (H) and the threshold J th :

[0044]

[0045] Among them, the scalar value H is the evaluation step size, defined as the 1-norm of the vector r(t) ∈ R l , r i (t) represents the i-th element of r(t), represents the supremum in the function value of J r (H).

[0046] Calculate the value of the residual evaluation function J r (H) within the evaluation step size according to the residual signal, and compare it with the set threshold J th . When J r (H) > J th , it is considered that the intelligent transportation signal system has a fault and is detected, and the system alarms; when J r (H) ≤ J th , it is considered that the intelligent transportation signal system has no fault.

[0047] The present invention has the following beneficial effects:

[0048] 1. A positive dynamic event-triggering mechanism is designed. Compared with the traditional time-triggering mechanism and static event-triggering mechanism, it can dynamically adjust the triggering conditions according to the changes in the traffic flow of the intelligent traffic signal system, reduce unnecessary data transmission, while ensuring the fault diagnosis effect, reduce the communication burden, improve the system response speed and stability, and can be applied to the actual complex traffic environment.

[0049] 2. Regarding the switching rules adopted for switched positive systems, the proposed MDMDT constraint is more in line with the switching characteristics of the intelligent traffic signal system. Based on this constraint, a new type of discrete linear co-positive Lyapunov function is constructed. Compared with the traditional Lyapunov function, by utilizing its affine dependence on the system matrix conditions and the combination with the non-negative vector v σ(t) (t), it is ensured that the function can still maintain a decreasing stability under the switching mode, so that the intelligent traffic signal system can operate smoothly, and the stability conditions can be transformed into a linear programming form, simplifying the solution process of the filter gain matrix and being more easily extended to systems with uncertainties and time-varying parameters.

[0050] 3. The designed fault diagnosis filter can provide a positive asymptotic estimate, which has both robustness to external disturbances and high sensitivity to faults, thus realizing smooth and effective fault diagnosis and ensuring the safety of urban traffic.

[0051] 4. By performing matrix decomposition on the filter parameter matrix, a mode-dependent piecewise fault diagnosis scheme based on linear programming is proposed. Compared with the existing linear matrix inequalities, it has a lower computational complexity and less conservatism. Description of the Drawings

[0052] Figure 1 is a schematic diagram of the fault diagnosis design framework of the switched positive system.

[0053] Figure 2 is a schematic diagram of the traffic signal system model at a triangular intersection.

[0054] Figure 3 is the switching signal σ(t) of the system.

[0055] Figure 4 is the curve trajectory diagram of the system state x(t) and its estimate .

[0056] Figure 5 are the release times and release intervals of the dynamic event trigger.

[0057] Figure 6 is the evolution process of the residual evaluation function J r (H). DETAILED DESCRIPTION

[0058] The present invention will be further explained below with reference to the accompanying drawings;

[0059] A multi-objective fault-disturbance collaborative diagnosis method for intelligent traffic signal systems, such as Figure 1 As shown in the figure, the sensor samples the intelligent traffic signal system and transmits the obtained measurement signal to the dynamic event trigger mechanism. The measurement signal that meets the predefined dynamic event trigger condition is transmitted through the network to the fault diagnosis filter based on MDMDT constraints to perform the corresponding fault diagnosis. The specific steps are as follows:

[0060] Step 1: Figure 2 As shown in the figure, the traffic lights in the intelligent traffic signal system are divided into two states: red and green lights, and the different stages of the red and green light states are switched according to the traffic flow dependent strategy. The following switching positive system state space model is established to describe the intelligent traffic signal system:

[0061]

[0062] Where x(t)∈R n represents the system status, n represents the number of buffers at the intersection, represents the first-order derivative of x(t), y(t)∈R p represents the measurement output, p represents the number of flow observation functions, w(t)∈R w represents the disturbance input due to sensor error and adjacent lane crossing, w represents the number of disturbances, f(t)∈R f represents the fault input in the intersection buffer, and f represents the number of faults. σ(t):[t0,∞)→ N ={1,2,…,N} is the switching signal of the switching system, N represents the number of subsystems, σ(t)=i∈ N Indicates that the i-th subsystem is activated, where N represents a finite set of subsystem numbers. x0 and t0 represent the initial state and initial time of the system respectively. A i ∈R n×n , D i ∈R n×w 、F i ∈R n×f , C i ∈R p×n is a known system matrix determined by the intersection structure.

[0063] This embodiment uses Figure 2Taking the shown triangular intersection as an example, which includes three roads A, B, and C, six traffic lights and three buffer zones. To improve the traffic efficiency, three symmetric signal configurations as shown in Table 1 are adopted, making the signals switch periodically between different configurations, with each stage lasting for a certain period of time to ensure that vehicles in all directions have sufficient passing time, and dynamically adjusting the signal switching time in combination with real-time traffic flow information:

[0064] Table 1

[0065] Intersection A Intersection B Intersection C <![CDATA[Buffer x1]]> <![CDATA[Buffer x2]]> <![CDATA[Buffer x3]]> Phase 1 Red Green Green Green Green Red Phase 2 Green Green Red Red Green Green Phase 3 Green Red Green Green Red Green

[0066] In a complex and changeable traffic environment, the operation of the system may be affected by various factors, such as sensor errors, the traffic flow entering the buffer zone caused by lane changes in adjacent lanes, and the irregular inflow into the buffer zone due to traffic light system failures. Also, since the number of vehicles waiting for traffic lights at the intersection, i.e., the system state, is always non-negative, it shows that the switched positive system established in Step 1 is very suitable for describing the intelligent traffic signal system.

[0067] Step 2: When the traffic flow at the intersection is too large, problems such as signal control imbalance and increased traffic delays will inevitably occur in the system. To avoid network communication delays, a dynamic event-triggered mechanism can be introduced to reduce unnecessary data processing and adjustment frequencies, and construct a fault diagnosis filter for the system under the dynamic event-triggered mechanism and MDMDT constraints.

[0068] s2.1: Design a fault diagnosis filter for the system under the dynamic event-triggered mechanism:

[0069]

[0070] Among them, represents the estimated vector of x(t), r(t) ∈ R l represents the residual signal, l represents the number of residual signals, is the unknown filter parameter matrix, is the sampled signal of the system at the h-th dynamic event trigger moment under, h ∈ Z, Z represents the set of natural numbers. Based on this, the following dynamic event-triggered mechanism is established:

[0071]

[0072] Among them, η(t) ≥ 0 represents the internal dynamic variable, θ > 0 represents the participation degree of η(t). As θ increases, the influence of η(t) on the system gradually decreases. ξ > 0 represents the attenuation rate of the filter. As ξ increases, the attenuation rate increases, meaning that fewer signal components are retained after filtering. 0 ≤ ψ i<1 represents the tolerance of the dynamic event-triggering mechanism. As ψ i increases, the tolerance for the error gradually increases, meaning that the amount of data transmitted will decrease. Ω i ∈R p represents an unknown dynamic event-triggering vector. The superscript T represents the transpose. inf{} represents the infimum.

[0073] Considering the positive constraint of the system, the sampling error is redefined as:

[0074]

[0075] By choosing an appropriate ε such that e y (t)≥0, the updated dynamic event-triggering mechanism is as follows:

[0076]

[0077] Under this dynamic event-triggering mechanism, define According to the switched positive system state-space model of the intelligent traffic signal system, the following augmented switched positive system is obtained:

[0078]

[0079] where 0 n×n 、0 n×p 、0 n×w 、0 n×f represent n×n, n×p, n×w, and n×f matrices with all elements being 0, respectively.

[0080] s2.2. Design the mode-dependent minimum dwell time MDMDT constraint:

[0081]

[0082] where T ki represents the dwell time during which the i-th subsystem is activated in the switching interval [t k ,t k+1 ), and represents the mode-dependent minimum dwell time during which the i-th subsystem is activated.

[0083] Based on the MDMDT constraint, the switching interval [t k ,t k+1 ) is divided into and two parts, and the interval is divided into S parts, with the length of each part being The s-th interval is denoted as

[0084] S2.3. Modify the fault diagnosis filter based on the dynamic event triggering mechanism in S2.1 and the MDMDT constraint in S2.2 as follows:

[0085]

[0086] Furthermore, a new augmented system can be obtained:

[0087]

[0088] where is mode-dependent and piecewise, and is also mode-dependent and piecewise.

[0089] Step 3. Design the conditions for the stable operation of the augmented dynamic system: Given parameters ε > 1, θ > 0, 0 ≤ ψ i < 1, ξ > 0, S ∈ Z + , If there exist constants γ > 0, β > 0, and vectors such that the following conditions hold:

[0090]

[0091]

[0092] then the augmented system in S2.3 is positive and exponentially stable under the MDMDT constraint and has an L - / L1 performance index, indicating that the fault diagnosis filter in S2.3 provides a positive asymptotic estimate with L - / L1 performance for the intelligent transportation signal system in Step 1. Where 0 p×n denotes a p-by-n matrix with all elements being 0, 1 l , 1 w , 1 f denote l-dimensional, w-dimensional, and f-dimensional column vectors with all elements being 1, respectively, denotes a Metzler matrix.

[0093] Step 4. The positive verification process of the augmented switched positive system is as follows:

[0094] It is known that the intelligent transportation signal system constructed in Step 1 satisfies Combined with the conditions in Step 3, we can obtain This indicates that Therefore, under any non - negative initial conditions, the positivity of the augmented switched positive system designed in S2.3 is guaranteed.

[0095] Step 5. The stability verification process of the augmented switched positive system is as follows:

[0096] Select the following discrete - type linear co - positive Lyapunov function:

[0097]

[0098] where is a time - dependent piece - wise linear function, denotes that the function values are positive real numbers and the number is 2n:

[0099]

[0100] where, denotes the value of the piece - wise function v i (t) at the critical point. Take the derivative of v i (t):

[0101]

[0102] Based on whether there is a dynamic event triggering moment within the subsystem switching interval [t k , t k+1 ), the exponential stability is discussed in two cases.

[0103] S5.1. For Consider two cases of the augmented system stability:

[0104] S5.1.1. Assume that the switching interval does not contain any dynamic event triggering moment, that is and Combined with the conditions in Step 3, it can be obtained that:

[0105]

[0106] Furthermore, when there is:

[0107]

[0108] S5.1.2. Assume that the switching interval contains several dynamic event triggering moments, that is Then for Repeat the steps in S5.1.1, and it can be obtained that:

[0109]

[0110] Similarly, for there exists:

[0111]

[0112] For there also exists:

[0113]

[0114] And so on, for each dynamic event trigger interval Repeating the steps in s5.1.1, conclusions similar to those in s5.1.1 can be obtained. Since Based on the conclusions obtained from each dynamic event trigger interval, for there are:

[0115]

[0116] Therefore, regardless of whether there is a dynamic event trigger moment in the switching interval the inequality always holds.

[0117] s5.2. For also consider two cases of the augmented system stability:

[0118] s5.2.1. Assume that the switching interval does not contain any dynamic event trigger moments, that is and Combining the conditions in step 3, it can be concluded that:

[0119]

[0120] Furthermore, when there are:

[0121]

[0122] s5.2.2. Assume that the switching interval contains several dynamic event trigger moments, that is Then for repeating the steps in s5.2.1, it can be obtained that:

[0123]

[0124] Similarly, for there are:

[0125]

[0126] For it also satisfies:

[0127]

[0128] For each dynamic event trigger interval Repeating the steps in s5.2.1, conclusions similar to those in s5.2.1 can be obtained. Since Based on the conclusions obtained from each dynamic event trigger interval, for there is:

[0129]

[0130] It can be obtained that regardless of whether there is a dynamic event trigger moment in the switching interval the inequality always holds.

[0131] Combining the discussion on the switching interval and for t ∈ [t k , t k+1 ), there is:

[0132]

[0133] Considering the switching moment t k , there exists σ(t) = i, t ∈ [t k , t k+1 ), and σ(t) = j, t ∈ [t k-1 , t k ). Combining the stability conditions in step 3, it can be concluded that:

[0134]

[0135] where and respectively represent the right limit and left limit of the switching moment t k . Furthermore, there is:

[0136]

[0137] Define where min{·} and max{·} respectively represent the minimum and maximum terms of the elements in the vector. It can be obtained that:

[0138]

[0139] Based on the above, there is:

[0140]

[0141] Therefore, the augmented switched positive system designed in s2.3 is exponentially stable.

[0142] Step 6: The verification process of the augmented switched positive system having L - / L1 performance is as follows:

[0143] Combining the conditions in Step 3, it can be obtained that Substitute into the conditions in Step 3 again, and it can be obtained that and That is and Namely and Furthermore, there is:

[0144]

[0145] It means that the augmented switched positive system designed in s2.3 has an L - / L1 performance index γ-β. Where is violation, that is, it satisfies I 2n represents the 2n×2n identity matrix, G wr (·) represents the transfer function from w(t) to r(t) when f(t)=0, G fr (·) represents the transfer function from f(t) to r(t) when w(t)=0, γ represents the L1 robust performance index of the residual signal r(t) to the disturbance w(t), β represents the L - sensitivity performance index of the residual signal r(t) to the fault f(t), sup represents the supremum in the function values, and inf represents the infimum in the function values.

[0146] Step 7: Solve the fault diagnosis filter gain matrix:

[0147] Given parameters ε>1, θ>0, 0≤ψ i <1, ξ>0, μ>0, S∈Z + , If there exist constants φ i >0, γ>0, β>0, and vector such that the following conditions hold:

[0148]

[0149] Then a positive - estimated and exponentially stable L - / L1 fault - diagnosis filter can be obtained. Where, I n represents an n - row and n - column identity matrix, and 1 n represents an n - dimensional column vector with all elements being 1, represents an n - dimensional column vector with the ι - th row element being 1 and the remaining elements being 0, that is, The filter parameter matrix is specifically designed as follows:

[0150]

[0151] Step 8: To illustrate the rationality of this method, theoretical verification is carried out on the fault - diagnosis filter designed above:

[0152] s8.1. The positive - property verification process of the fault - diagnosis filter is as follows:

[0153] Since it can be obtained that Combined with the conditions in Step 7, there is:

[0154]

[0155] According to the specific form of the fault - diagnosis filter parameter matrix in Step 7, it can be concluded that:

[0156]

[0157] That is, is a Metzler matrix, that is, Similarly, since Combined with the specific form of the fault - diagnosis filter parameter matrix in Step 7, it can be obtained that Similarly, since Combined with the conditions in Step 7 and the specific forms of the fault - diagnosis filter parameter matrices and it can be obtained that Therefore, the designed fault - diagnosis filter can provide a positive estimate.

[0158] s8.2. The stability and L - / L1 performance verification process is as follows:

[0159] According to the conditions in Step 7, there is:

[0160]

[0161] Based on the above, the conditions in step 7 are equivalent to:

[0162]

[0163] Combined with the conditions in step 7, we have:

[0164]

[0165] Furthermore, we can obtain:

[0166]

[0167] Then we have:

[0168]

[0169] Corresponding to the conditions in step 3, similarly, the other conditions in step 7 can also be obtained by this method. Among them,

[0170] Therefore, when the conditions in step 7 are satisfied, the designed fault diagnosis filter can provide a positive estimate with L - / L1 performance under the exponential stability condition.

[0171] s8.3. Discussion on Zeno behavior:

[0172] The designed fault diagnosis filter in s2.3 should not only ensure the stability of the augmented dynamic system but also check the triggering behavior. To avoid Zeno behavior, that is, an infinite number of event triggers occur within a finite time interval, a positive lower bound τ of the time interval between any two consecutive samplings is given.

[0173] According to the sampling error in s2.1 It can be obtained that:

[0174]

[0175] Among them, h i = ||A i ||1, Solve the differential equation And Then we have:

[0176]

[0177] Combined with the dynamic event-triggering mechanism in s2.1, we can obtain:

[0178]

[0179] Furthermore, there is:

[0180]

[0181] Thus, we obtain:

[0182]

[0183] Based on the above, define The lower bound τ of the time interval can be obtained as: is:

[0184]

[0185] Wherein,

[0186] Step 9: To verify the effectiveness of this method, the designed fault diagnosis filter will be simulated and verified using MATLAB software below.

[0187] Assume that the number of lanes on each road at the intersection is different. Specifically, Road A has 2 lanes, Road B has 3 lanes, and Road C has 1 lane. Set the system matrix:

[0188]

[0189] Wherein, A i represents the dynamic characteristics of the traffic flow in the buffer zone, C i represents the measurement matrix of the traffic flow monitoring device, D i represents the disturbance effect of the disturbance input on the traffic flow in the buffer zone, F i represents the influence of the fault input on the traffic flow in the buffer zone, And set the MDMDT constraint as:

[0190]

[0191] Note that in the intelligent traffic signal system, the MDMDT constraint T ki is determined by the actual traffic flow x(t) at the current intersection and can be further expressed as:

[0192]

[0193] Set the parameters of the dynamic event-triggering mechanism as ε = 1.2, θ = 0.1, ψ1 = 0.4, ψ2 = 0.6, ψ3 = 0.8, ξ = 1, and other parameters μ = 1, S = 2, select the appropriate vector Solving the conditions in Step 7 through the feasp solver can obtain an L -The performance index of / L1 is γ-β = 9.7975 - 6.1416 = 3.6559, and the filter gain matrix is:

[0194]

[0195]

[0196] The dynamic event trigger vector is:

[0197]

[0198] Set the initial state vector of the system as x(0) = [10 10 10] T , and the initial estimated state vector of the filter is The disturbance input is w(t) = [0.05e -0.5t 0.04|cos5t| 0.06|2t - 1|] T , and the fault input is:

[0199]

[0200] where, u i (t) is an indicator function. When the i-th fault is activated, u i (t) = 1; otherwise, u i (t) = 0. Obviously, u i (t) satisfies f1(t) = 0.5t - 2.5, f2(t) = 0.4t - 0.5, f3(t) = 0.3t - 1.5.

[0201] Through MATLAB simulation, the switching signal that satisfies the MDMDT constraint can be obtained as Figure 3 shown. The curve trajectories of the system state and its estimation are as Figure 4 shown. The release moments and release intervals of the dynamic event trigger are as Figure 5 shown. It can be seen that the number of trigger times of the system within a finite time is 27 times. The evolution process of the residual evaluation function J r (H) is as Figure 6 shown, where the solid line corresponds to the case with faults, and the dashed line corresponds to the case without faults. When t = 11.2 s, the threshold of the system can be set as J th = 12.0978. The simulation results show that when t = 11.6 s, J r (H) = 12.0993 > 12.0978, which means that after the fault signal occurs at t = 10 s, it can be detected by the fault diagnosis filter after 1.6 s, indicating the effectiveness of the fault diagnosis filter designed by the present invention.

Claims

1. A multi-objective fault-disturbance collaborative diagnosis method for an intelligent traffic signal system, characterized in that: The specific steps include: Step 1: Taking the traffic flow in the intersection buffer as the system state x(t), taking the traffic flow observation function value as the measurement output y(t), considering the disturbance input w(t) and the fault input f(t) in the intersection buffer, a switching positive system state space model of the intelligent traffic signal system is established; Step 2: Design a traffic signal system fault diagnosis filter under the dynamic event trigger mechanism and the modal-dependent minimum dwell time MDMDT constraint; Step 3: Use linear programming method to design and optimize the conditions for the smooth operation of the augmented switching positive system; Step 4: Solve the fault diagnosis filter parameter matrix and calculate the residual evaluation function J r (H) and threshold J th : Among them, the scalar value H is the evaluation step size, ||r(t)||1 is the 1-norm of the residual signal r(t), Indicates J r (H) The supremum of the function value, f(t) represents the fault input in the intersection buffer; when J r (H)>J th When the intelligent traffic signal system is considered to have failed and has been detected, the system will alarm; when J r (H)≤J th It is considered that the intelligent traffic signal system has no fault.

2. A multi-objective fault-disturbance collaborative diagnosis method for an intelligent traffic signal system as claimed in claim 1, characterized in that: The switching positive system state space model of the intelligent traffic signal system is: Where x(t)∈R n , n represents the number of buffer zones at the intersection, represents the first-order derivative of x(t); y(t)∈R p , p represents the number of flow observation functions; w(t)∈R w represents the disturbance input caused by sensor error and adjacent lane crossing, w represents the number of disturbances; f(t)∈R f represents the fault input in the intersection buffer, f represents the number of faults; σ(t):[t0,∞)→ N ={1,2,…,N} is the switching signal, σ(t)=i∈ N Indicates that the i-th subsystem is activated, where N represents a finite set of subsystem numbers, N represents the number of subsystems; x0 and t0 represent the initial state and initial time of the system respectively; A i ∈R n×n , D i ∈R n×w 、F i ∈R n×f , C i ∈R p×n is a known system matrix determined by the intersection structure.

3. A multi-objective fault-disturbance collaborative diagnosis method for an intelligent traffic signal system as claimed in claim 1, characterized in that: Design fault diagnosis filter for the system under dynamic event trigger mechanism: in, represents the estimated vector of x(t), r(t)∈R l represents the residual signal, l represents the number of residual signals, is the unknown filter parameter matrix, N A finite set representing the number of subsystems; is the time when the system triggers the hth dynamic event The sampled signal under , h∈Z, Z represents a set of natural numbers.

4. A multi-objective fault-disturbance collaborative diagnosis method for an intelligent traffic signal system as claimed in claim 3, characterized in that: The dynamic event triggering mechanism is designed as follows: Among them, η(t)≥0 represents the internal dynamic variable, θ>0 represents the degree of participation of η(t); ξ>0 represents the attenuation rate of the filter; 0≤ψ i <1 indicates the tolerance of the dynamic event trigger mechanism; Ω σ(t) represents the unknown dynamic event trigger vector, σ(t)=i∈ N Indicates that the i-th subsystem is activated; superscript T indicates transposition; inf{} indicates the infimum; Define the sampling error e y (t) is: Choose ε so that e y (t)≥0, update the dynamic event trigger mechanism to: Introduce the mode-dependent minimum dwell time MDMDT constraint and change the switching interval [t k ,t k+1 ) is divided into and Two parts, the interval Divided into S parts, represents the modal-dependent minimum dwell time for the i-th subsystem to be activated, and the s-th interval is expressed as Modify the fault diagnostic filter to: definition The following augmented system is obtained: in, is modality-dependent and piecewise.

5. A multi-objective fault-disturbance collaborative diagnosis method for an intelligent traffic signal system as claimed in claim 4, characterized in that: Choose the following discrete linear copositive Lyapunov function: in is a time-dependent piecewise linear function, Indicates that the function value is a positive real number and its quantity is 2n: in, Represents the piecewise function v i (t) value at the critical point; Design the conditions for the stable operation of the augmented dynamic system, and analyze the positivity, stability and L of the augmented system - / L1 performance conditions, design the parameter matrix of the fault diagnosis filter for: Among them, the constant φ i >0, vector I n represents the identity matrix of n rows and n columns, 1 n represents an n-dimensional column vector whose elements are all 1, represents an n-dimensional column vector whose ιth row element is 1 and the rest of the elements are 0, that is, Superscript Represents the inverse of the matrix.

6. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the method according to any one of claims 1 to 5.

Images