Ship and bridge collision detection method and system, computer equipment and storage medium

Abstract

The invention provides a ship-bridge collision detection method and system, computer equipment and a storage medium, and belongs to the field of bridge structure health monitoring and safety alarming, and the method comprises the steps: collecting the displacement response of a bridge structure key position under ship collision disturbance; the method comprises the following steps: firstly, expanding acquired response data in a frequency domain by using discrete wavelet transform, extracting a low-frequency approximate signal to carry out co-integration test to obtain a multi-scale co-integration equation, and carrying out online real-time estimation on a co-integration coefficient in each scale equation by using a Kalman filtering recursive algorithm; when the co-integration coefficient changes and exceeds a safety limit, automatic alarm is realized; alarm results of co-integration coefficients of multiple scales are synthesized, and when 75% of co-integration combined alarm is set to be exceeded, a ship and bridge collision alarm decision is made. According to the method, the Kalman filtering algorithm is combined, so that the change of the co-integration coefficient can be predicted on line, and the actual engineering requirements are met. On the basis, multi-scale wavelet decomposition and co-integration improve the robustness and precision of alarm, a plurality of alarm results are provided for comprehensive decision making, and scientific basis and technical support are provided for bridge structure health monitoring and accident prevention and control.

Description

Technical Field

[0001] This invention belongs to the field of bridge structural health monitoring and safety alarm technology, specifically relating to a ship-bridge collision detection method, system, computer equipment, and storage medium. Background Technology

[0002] In recent years, my country's inland waterway shipping industry has experienced rapid development, with a continuous increase in waterway traffic density. Consequently, ship-bridge collisions have become increasingly frequent. These accidents not only cause structural damage to bridges, but in severe cases, can even lead to structural collapse, directly threatening bridge operational safety and the smooth flow of public transportation. More alarmingly, if alarms are not issued promptly after an accident, it can easily trigger secondary disasters such as vehicles falling off bridges and secondary collisions, further increasing casualties and property damage. Therefore, developing efficient ship-bridge collision monitoring and alarm methods to improve the timeliness and accuracy of accident identification is of significant engineering application value and practical importance for extending the service life of bridges, ensuring traffic safety, and providing emergency response technical support for bridge maintenance departments.

[0003] Currently, research on ship-bridge collisions largely focuses on developing early warning systems before accidents occur and analyzing impact force changes after accidents. However, there are still significant shortcomings in the accurate identification of collision behavior itself, especially in research on real-time alarm systems to prevent secondary disasters. Although some existing bridges are equipped with ship-bridge collision early warning systems, they face many practical challenges in actual operation: extreme weather (such as heavy fog and rainstorms) can interfere with the normal operation of the equipment, and hardware failures or software malfunctions can reduce system reliability, leading to frequent accidents where ships breach the early warning system and ultimately collide with the bridge. Therefore, developing an efficient alarm method based on bridge structural response signals to improve the monitoring and identification capabilities of ship-bridge collision accidents has significant engineering application value and practical significance for extending the service life of bridges, ensuring traffic safety, and providing timely and effective emergency response technical support for bridge maintenance departments.

[0004] However, current research on alarm methods for ship-bridge collision accidents is relatively limited. It mainly relies on known systems and their power output signals, utilizing response data collected by monitoring sensors, and employing time-domain analysis, frequency-domain analysis, and other methods to identify abnormal bridge behavior. The main research methods are as follows: Time-domain identification methods mainly include threshold discrimination and rate of change detection. Threshold discrimination involves setting alarm thresholds for indicators such as displacement, acceleration, and strain; when the response exceeds the threshold, an alarm is triggered immediately. Its advantages are simplicity and ease of real-time implementation, but it is prone to false alarms due to wind vibration, traffic, and other interference. Rate of change detection monitors the rate of change of the response quantity, enabling more sensitive detection of sudden impacts, but still requires reasonable threshold settings. Frequency-domain analysis methods mainly include high-frequency energy mutation and characteristic frequency drift detection. High-frequency energy mutation triggers an alarm based on drastic changes in high-frequency energy during impact, exhibiting extremely high sensitivity, but is easily affected by other high-frequency noise. Characteristic frequency drift detection indicates structural damage by tracking minute changes in the bridge's natural frequency, suitable for subsequent damage assessment, but has limited immediate identification capability for initial impacts. Joint time-frequency analysis, such as wavelet packet energy distribution anomaly detection, considers both time and frequency characteristics, enabling more accurate extraction of impact signals, but has high computational complexity, making it suitable for post-processing or high-performance computing platforms. In recent years, data-driven methods (such as support vector machines and neural networks) have also been introduced into alarm systems. By learning the characteristics of normal states, they can identify anomalies in real time and have good adaptability and noise resistance. However, they require a large amount of normal data for training, and the excessive computation time also leads to insufficient real-time performance.

[0005] Overall, while existing technologies have gradually improved in detection accuracy and anti-interference capabilities, they have yet to achieve a balance between high robustness, high real-time performance, and high precision. Time-frequency joint analysis methods are limited by real-time requirements, and data-driven methods suffer from data dependence and time consumption. These technological shortcomings mean the industry lacks a core solution capable of accurately estimating parameters and efficiently capturing sudden parameter changes, becoming a key bottleneck restricting the prevention and control of secondary disasters from ship-bridge collisions and improving bridge operational safety. Summary of the Invention

[0006] To address the aforementioned problems, this invention provides a method for detecting collisions between ships and bridges.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A bridge collision detection method includes the following steps: The original displacement response data of the two lateral sides of the impacted bridge pier at the same height were collected. The two sets of displacement response data were subjected to L-level discrete wavelet transform to decompose the multi-scale approximate signals of each layer of the two sets of displacement response. The stationarity of the L-level multi-scale approximate signals of the two sets of displacement response data is tested, and the Johansen cointegration test is performed on the selected non-stationary signals of the same scale on both sides, resulting in L cointegration equations. Each cointegration equation is treated as a state equation, and the cointegration coefficients of the cointegration equations constitute the initial state vector. xThe multi-scale approximation signals at each level are used as observation variables, and the Kalman filter algorithm is used to calculate and obtain the estimated values ​​of the cointegration coefficients at each scale in real time. The estimated values ​​of the cointegration coefficients at the current time at each scale are compared with the set safety limits. When the estimated values ​​of the cointegration coefficients at the current time at three or more scales exceed the safety limits, a ship-bridge collision alarm is triggered, and a final emergency alarm signal is pushed.

[0008] Preferably, a low-pass filter is used to perform multi-level discrete wavelet transform on the displacement response data to decompose it into multi-scale approximate signals of each level of the displacement response. The specific formula is as follows: ; In the formula, It is the first Multi-scale approximation signal of the layer; These are the coefficients of a low-pass filter, used to extract low-frequency components; n This refers to the number of sampling points for the displacement response signal; It is the filter length. k = n .

[0009] Preferably, the test model for the stationarity test of the L-level multi-scale approximate signal of the two sets of displacement response data is as follows: ; In the formula, The original displacement response data, For time indexing, The position index of the lag order, For constant terms, This is the trend coefficient. The unit root coefficient, These are the model coefficients. Defined as , It is a white noise process. It is the lag order of the autoregressive process; After being identified as a non-stationary sequence through stationarity testing, Johansen cointegration tests were performed on the non-stationary signals of the same scale on both sides, yielding multiple cointegration equations: ; ; | ; In the formula, and These are multi-scale approximate signals at the same height on both sides of the bridge pier, representing different layers. and These are all approximate signals corresponding to the scale decomposition of the original displacement response data. and For the corresponding cointegration coefficient, This is a random disturbance term.

[0010] Preferably, each cointegration equation is used as a state equation, and the cointegration coefficients of the cointegration equations constitute the initial state vector. x Using the multi-scale approximation signals at each level as observation variables, the Kalman filter algorithm is used to calculate and obtain the estimated values ​​of the cointegration coefficients at each scale in real time. Specifically: ; ; In the formula, It is represented as a state vector composed of cointegration coefficients. Refers to the state vector rate of change over time The cointegration coefficient; This represents the noise of a zero-mean stationary process with covariance Q. For observation noise with covariance matrix R, the observation matrix is ​​represented as: ; The estimated values ​​of the cointegration coefficients at each scale are obtained in real time through prediction updates and measurement updates, as detailed below: Forecast Update: ; ; Measurement Update: ; ; ; In the formula, for The observation matrix at time, for The transpose of the observation matrix at each time step. yes The state vector at time t, yes Moment The estimated cointegration coefficients, yes time The predicted value of the cointegration coefficient, , They are , The relevant error covariance matrix, It is the identity matrix. It is an observed variable. It is the Kalman gain matrix. It is the error covariance matrix at time k. Time-based process noise, Observational noise at any given moment.

[0011] Preferably, a fixed safety limit is set using a statistical control process method; when the estimated value of the cointegration coefficient remains within the safety limit, no abnormal alarm is triggered, and when it exceeds the safety limit, an automatic alarm is triggered; when the alarm results of the four-scale cointegration combination are combined, a bridge collision alarm decision is made when three or more cointegration combination alarms are set.

[0012] Preferably, the fixed safety limit is set using a statistical control process method, and the specific formula for setting it is as follows: ; In the formula, These are the upper and lower limits of the safety boundary, respectively. This represents the average value of the estimated cointegration coefficients; This corresponds to the standard normal distribution value at a 95% confidence level; This represents the standard deviation of the estimated cointegration coefficient; This refers to the sample size.

[0013] The present invention also proposes a ship-bridge collision detection system, comprising: The data acquisition module is used to collect the original displacement response data at the same height on both sides of the impacted bridge pier. The two sets of displacement response data are subjected to L-level discrete wavelet transform to decompose the two sets of displacement response data into multi-scale approximate signals of each level. The data processing module is used to perform stationarity tests on the L-level multi-scale approximate signals of two sets of displacement response data, and to perform Johansen cointegration tests on the selected non-stationary signals of the same scale on both sides to obtain L cointegration equations. Each cointegration equation is treated as a state equation, and the cointegration coefficients of the cointegration equations constitute the initial state vector. x The multi-scale approximation signals at each level are used as observation variables, and the Kalman filter algorithm is used to calculate and obtain the estimated values ​​of the cointegration coefficients at each scale in real time. The alarm module is used to compare the estimated values ​​of the cointegration coefficients at the current time at each scale with the set safety limits. When the estimated values ​​of the cointegration coefficients at the current time at three or more scales exceed the safety limits, a ship-bridge collision alarm is triggered, and a final emergency alarm signal is pushed.

[0014] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement any of the steps in the bridge collision detection method.

[0015] The present invention also provides a computer-readable storage medium storing a computer program that, when loaded by a processor, can execute any of the steps in the bridge collision detection method.

[0016] The ship-bridge collision detection method provided by this invention has the following beneficial effects: This invention first identifies symmetrical monitoring points on both sides of the impacted bridge pier. Displacement response data is decomposed using multi-layer discrete wavelet transform to extract low-frequency approximate signals, filtering out interference from wind vibration and high-frequency noise, thus improving anti-interference capabilities. Next, four cointegration equations are constructed using stationarity and Johansen cointegration tests, serving as state equations. These equations, combined with parameters such as the initial state vector of the cointegration coefficients and process noise, are then incorporated into a Kalman filter. Real-time iterative cointegration coefficient estimates are obtained through prediction and measurement updates, eliminating the need for extensive data training and balancing computational efficiency and estimation accuracy. This overcomes the shortcomings of insufficient real-time performance in time-frequency joint analysis and the time-consuming nature of data-driven methods. Finally, a safety limit is set, and alarm results from multiple scales are integrated to achieve accurate collision behavior identification and real-time alarm. Through the core logic of multi-scale response fusion and state estimation-driven approaches, this invention precisely addresses the difficulty of simultaneously achieving robustness, real-time performance, and accuracy in existing technologies. Ultimately, it achieves a unity of high robustness, high real-time performance, and high accuracy, effectively preventing secondary disasters, providing timely and reliable emergency response support for bridge maintenance departments, and significantly improving the safety assurance capabilities of bridge operations. Attached Figure Description

[0017] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the ship-bridge collision detection method proposed in this invention.

[0019] Figure 2 This is a numerical model diagram of a barge-bridge collision in Embodiment 1 of the present invention; Figure 3 The original displacement response variable in Embodiment 1 of the present invention ; Figure 4 The original displacement response variable in Embodiment 1 of the present invention ; Figure 5The cointegration coefficient in Embodiment 1 of the present invention Estimated alarm results; Figure 6 The cointegration coefficient in Embodiment 1 of the present invention Estimated alarm result. Detailed Implementation

[0020] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0021] This invention proposes a ship-bridge collision detection method, specifically a real-time alarm method for ship-bridge collision accidents driven by multi-scale response fusion and state estimation. Based on historical measurements of cointegration features in the structural response signal under multi-scale decomposition, this method is used to estimate the cointegration coefficients and their evolution rules in real time to detect any abnormal changes in the selected features, thereby achieving timely identification and alarm of ship-bridge collision events. Figure 1 As shown, the specific steps for implementing this method include:

[0022] Step 1: Preliminary preparation and parameter setting.

[0023] The monitoring target of this invention is the bridge pier that is impacted in the ship-bridge collision system. It locks the same height position on both sides of the lateral side as the core monitoring point to ensure data symmetry and correlation.

[0024] Next, key parameters were set, including setting the number of discrete wavelet transform decomposition layers L, taking into account both feature extraction accuracy and computational efficiency. The invention sets the number of discrete wavelet transform layers L=4, the statistical confidence level to 95%, corresponding to the standard normal distribution value Z=1.96; the filter length and the number of displacement response signal acquisition points n are kept consistent.

[0025] Finally, equipment debugging and calibration are performed to ensure that the displacement sensor works normally, eliminate hardware fault interference, and ensure the accuracy and continuity of the collected data.

[0026] Step 2: Displacement response data acquisition and multi-scale decomposition.

[0027] Step 21: Synchronously collect two sets of displacement response data from monitoring points at the same height on both sides of the target pier, and record the number of collection points n and the signal acquisition time sequence.

[0028] Step 22: Perform a 4-level discrete wavelet transform on both sets of displacement response data to extract the multi-scale approximate signal at each level, i.e., the low-frequency component, denoted as . , , , The calculation formulas for the multi-scale approximation signals at each layer are as follows: (1) In the formula: For the first i Multi-scale approximation signal, i=1,2,3,4; These are the coefficients of a low-pass filter, used to extract low-frequency components; n This refers to the number of sampling points for the displacement response signal; It is the filter length. k = n .

[0029] This invention uses multi-scale wavelet transform to extract key features of bridge structural response at different time scales, enhancing adaptability to disturbances of different amplitudes and frequencies.

[0030] Step 3: The stationarity of the four-level multi-scale approximation signals of the two sets of displacement response data is tested. After filtering out non-stationary signals, a Johansen cointegration test is performed on the non-stationary signals at the same scale on both sides, resulting in four cointegration equations. This quantifies the long-term equilibrium relationship of the displacement responses on both sides of the bridge pier, eliminating spurious correlations caused by wind vibration, traffic, and other disturbances. The complex original signals are transformed into cointegration relationships with clear physical meaning, providing reliable state equations for real-time estimation of cointegration coefficients using Kalman filtering. This lays the foundation for subsequent accurate estimation of cointegration coefficients and the detection of signal anomalies caused by collisions.

[0031] As a further preferred embodiment of the present invention, the specific calculation method for step 3 is as follows: Cointegration theory is an analytical method for non-stationary sequences, so the stationarity of the multi-scale approximate signal obtained by discrete wavelet transform must first be tested. The test model is shown below: (2) In the formula, The original displacement response data, For constant terms, This is the trend coefficient. For time indexing, The position index is the lag order. The unit root coefficient, These are the model coefficients. Defined as , It is a white noise process. It is the lag order of the autoregressive process.

[0032] After identifying the sequence as non-stationary through stationarity testing, cointegration testing was performed to obtain multiple cointegration equations: ; ; | ; In the formula, and These are multi-scale approximate signals at the same height on both sides of the bridge pier, representing different layers. and These are all approximate signals corresponding to the scale decomposition of the original displacement response data. and These are the corresponding linear combination coefficients, also known as cointegration coefficients. This is a random disturbance term.

[0033] Step 4: Treat each cointegration equation as a state equation, and the cointegration coefficients form the initial state vector. x The multi-scale approximation signals at each layer are used as observation variables; process noise is set. Observation noise and observation matrix H The initial state vector is formed by the cointegration equation itself and the cointegration coefficients. x Observed variables, process noise Observation noise and observation matrix H These values ​​are then incorporated into the Kalman filter algorithm to obtain real-time estimates of the cointegration coefficients at various scales through prediction and measurement updates. The prediction update, based on the previous time step's cointegration coefficient estimate, the rate of change of the state vector, and process noise, predicts the current time step's cointegration coefficient and error covariance matrix—this is a priori estimation. The measurement update, combining the current time step's displacement response observation data, observation matrix, and observation noise, corrects the prediction update result to obtain a more accurate estimate of the current time step's cointegration coefficients—this is a posterior correction.

[0034] As a further preferred embodiment of the present invention, step 4 is implemented as follows: (6) (7) In the formula, It is represented as a state vector composed of cointegration coefficients. Refers to the state vector rate of change over time These are the cointegration coefficients. Because the initial state vector... Since the constant does not conform to time-varying characteristics, zero-mean white noise that takes into account environmental influences is introduced here. , indicating that the state vector change is related to it. This represents the noise of a zero-mean stationary process with covariance Q. For observation noise with covariance matrix R, the observation matrix can be represented as: .

[0035] Forecast Update: (7) (8) Measurement Update: (9) (10) (11) In the formula, for The observation matrix at time, for The transpose of the observation matrix at each time step. yes The state vector at time t, yes Moment The estimated cointegration coefficients, yes time The predicted value of the cointegration coefficient, , They are , The relevant error covariance matrix, It is the identity matrix. Observed variables It is the Kalman gain matrix. It is the error covariance matrix at time k. Time-based process noise, Observational noise at any given moment.

[0036] By utilizing cointegration theory, the long-term stable relationship between the response signals of multiple sensors is revealed, and the cointegration coefficient is used as a health status indicator to sensitively reflect abnormal structural changes. Combined with Kalman filtering technology, dynamic real-time estimation and iterative updating of the cointegration coefficient are achieved, thereby effectively capturing abrupt behavior and realizing accurate real-time alarm for ship-bridge collision accidents.

[0037] Step 5: Set fixed safety limits using statistical process control methods. LCL , UCLThe system monitors the estimated values ​​of cointegration coefficients at various scales in real time. No alarm is triggered when the estimated cointegration coefficients remain within safe limits; an automatic alarm is triggered when the estimated cointegration coefficients exceed safe limits. By combining the alarm results of the four-scale cointegration combinations, three or more cointegration combination alarms are set, and a final alarm for a ship-bridge collision accident is immediately executed, sending an emergency response signal to the bridge maintenance department.

[0038] As a further preferred embodiment of the present invention, step 5 is implemented as follows: Statistical process control limits are used as safety boundaries, and the sigma rule is applied with a defined applicable range to ensure a 95% confidence level. Therefore, the range near the normal value is defined as the adaptive statistical process control limits. Fixed safety boundaries are set using statistical control process methods, with the specific formula as follows:

[0039] (12) In the formula, These are the upper and lower limits of the safety boundary, respectively. This represents the average of the estimated cointegration coefficients; This corresponds to the standard normal distribution value at a 95% confidence level (take...). =1.96). This represents the standard deviation of the estimated cointegration coefficients; This refers to the sample size.

[0040] The advantages of the above technical solution adopted in this invention are: Unlike time-domain analysis and data-driven methods, which rely on raw sensor data and involve significant computational loads leading to low robustness and poor real-time performance, this invention presents a high-precision bridge collision alarm method with real-time parameter estimation and abrupt change identification capabilities, based on multi-scale response cointegration analysis and Kalman filtering. Through multi-scale wavelet transform, key features in the bridge structure response can be extracted at different time scales, enhancing adaptability to disturbances of varying amplitudes and frequencies. Cointegration theory reveals the long-term stable relationship between multi-sensor response signals, and the cointegration coefficient serves as a health indicator, sensitively reflecting abnormal structural changes. Combined with Kalman filtering, dynamic real-time estimation and iterative updates of the cointegration coefficient are achieved, effectively capturing abrupt changes. Simultaneously, this invention introduces the sigma rule and defines its applicable range to ensure a 95% confidence level. Therefore, the range near the normal value is defined as the statistical process control safety limit.

[0041] Compared to traditional structural health monitoring methods based on cointegration theory, this invention combines a Kalman filter recursive algorithm, enabling online prediction of changes in cointegration coefficients, thus meeting practical engineering needs. Furthermore, multi-scale wavelet decomposition and cointegration improve the robustness and accuracy of alarms, and provide multiple alarm results for comprehensive decision-making. All of these provide a scientific basis and technical support for bridge structural health monitoring and accident prevention.

[0042] Example 1 The following is a detailed description of the specific implementation methods in conjunction with specific embodiments.

[0043] The example is a numerical simulation of a two-span continuous beam bridge. The detailed calculation steps are as follows: A refined numerical model of the barge-beam bridge collision was developed, and its displacement response matches the actual situation. The numerical model and sensor locations are shown below. Figure 2 .

[0044] Step 1: Apply random wind loads to a finite element model of a two-span continuous beam bridge to simulate the bridge's normal operating environment. Simulate a scenario where the middle pier of the bridge is struck by a barge. Extract the original displacement response data at the same height on both sides of the pier top under the combined effects of wind load and barge impact. Figure 3 and Figure 4 .

[0045] Step 2: Import the data into the alarm method, setting the collision between the ship and bridge to occur at 1400 ms. Multi-scale wavelet decomposition effectively preserves the key features of the original signal, especially the overall deformation and long-term trend of the low-frequency component, which is crucial for capturing sudden collision events. By updating the cointegration coefficients in real time through cointegration analysis and Kalman filtering, a significant abrupt change occurs at 1400 ms, exceeding the safety limit and accurately triggering the alarm. Alarms were successfully triggered at all four scales. Comprehensive decision-making shows that the combination of multi-scale wavelet decomposition and cointegration analysis significantly improves alarm accuracy. Controlling the wavelet decomposition to levels 1 to 4 allows for the extraction of key features and identification of collision events while avoiding interference from redundant low-frequency information, thus enhancing the robustness and response speed of the method under complex operating conditions. See the multi-scale alarm effect diagram below. Figure 5 and Figure 6 .

[0046] The alarm effect diagram shows that the system can effectively track changes in the cointegration coefficient, achieving rapid convergence and dynamic capture within a short time. When changes exceed safety limits, they can be detected promptly.

[0047] This invention obtains multi-scale approximate signals by performing discrete wavelet transform on the displacement responses of both sides of bridge piers. Then, it constructs cointegration equations by combining stationarity tests and Johansen cointegration tests. This effectively filters environmental interference such as wind vibration and traffic noise, improving the anti-interference capability of signal extraction and enhancing the robustness of detection. It solves the problem of insufficient robustness in basic time-domain and frequency-domain analysis methods.

[0048] Furthermore, this invention introduces a Kalman filter algorithm to estimate the cointegration coefficients in real time. Compared to the complex calculations of the wavelet packet energy distribution anomaly detection method, the process is more targeted, balancing the accuracy of time-frequency domain feature extraction with the efficiency of algorithm execution, thus meeting the requirements of real-time detection. This solves the problem of high computational complexity and difficulty in adapting to real-time detection of conventional bridges using joint time-frequency analysis methods.

[0049] Building upon this foundation, this invention eliminates the need for training models with extensive normal operating condition data. Instead, it implements alarms based on the cointegration relationship and statistical control limits of displacement response, significantly reducing data dependence. Furthermore, by employing multi-scale monitoring and comprehensive decision-making based on multi-scale results, it improves the accuracy of collision identification, avoiding false alarms and missed alarms caused by single-scale judgments. This solves the problems of data-driven methods relying on massive amounts of training data and insufficient real-time response capabilities.

[0050] In summary, this invention overcomes the bottleneck of existing technologies in achieving robustness, real-time performance, and accuracy by combining multi-scale signal processing, cointegration analysis, Kalman filtering, and statistical decision-making. It solves the industry pain point of lacking core technologies that can accurately estimate parameters and efficiently capture parameter mutations, thus failing to achieve real-time and accurate alarm for ship-bridge collisions.

[0051] Based on the same inventive concept, this invention provides a ship-bridge collision detection system, including a data acquisition module, a data processing module, and an alarm module: Specifically, the data acquisition module is used to collect the original displacement response data at the same height on both sides of the impacted bridge pier, and to perform L-level discrete wavelet transform on the two sets of displacement response data to decompose them into multi-scale approximate signals of each layer of the two sets of displacement responses.

[0052] The data processing module is used to perform stationarity tests on the L-level multi-scale approximate signals of the two sets of displacement response data, and to perform Johansen cointegration tests on the selected non-stationary signals of the same scale on both sides, resulting in L cointegration equations.

[0053] Each cointegration equation is used as a state equation, and the cointegration coefficients of the cointegration equations form the initial state vector. x The multi-scale approximation signals at each level are used as observation variables, and the Kalman filter algorithm is used to calculate and obtain the estimated values ​​of the cointegration coefficients at each scale in real time.

[0054] The alarm module compares the estimated values ​​of the cointegration coefficients at the current time at each scale with the set safety limits. When the estimated values ​​of the cointegration coefficients at the current time at three or more scales exceed the safety limits, a ship-bridge collision alarm is triggered, and a final emergency alarm signal is pushed.

[0055] Each module in the aforementioned ship-bridge collision detection system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0056] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps in the embodiments of the ship-bridge collision detection method. Specific implementation methods can be found in the method embodiments, and will not be repeated here.

[0057] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions, on which a computer program is stored. For example, a memory containing instructions that can be executed by a processor of a computer device to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. When the computer program is executed by the processor, it can implement the steps in the embodiments of the bridge collision detection method. Specific implementation methods can be found in the method embodiments, which will not be repeated here.

[0058] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0059] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0060] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0061] These computer program instructions can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0062] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail in this specification and embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered within the protection scope of the present invention patent. No reference numerals in the claims should be construed as limiting the scope of the claims. Any simple variations or equivalent substitutions of technical solutions that can be readily obtained by those skilled in the art within the scope of the technology disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A method of ship bridge collision detection, characterized by, include: The original displacement response data of the two lateral sides of the impacted bridge pier at the same height were collected. The two sets of displacement response data were subjected to L-level discrete wavelet transform to decompose the multi-scale approximate signals of each layer of the two sets of displacement response. The stationarity of the L-level multi-scale approximate signals of the two sets of displacement response data is tested, and the Johansen cointegration test is performed on the selected non-stationary signals of the same scale on both sides, resulting in L cointegration equations. Each cointegration equation is taken as a state equation, and the cointegration coefficients of the cointegration equation constitute an initial state vector x Each layer multi-scale approximation signal is taken as an observation variable, and the Kalman filtering algorithm is used for calculation to obtain the cointegration coefficient estimation value of each scale at the current time in real time. The estimated values ​​of the cointegration coefficients at the current time at each scale are compared with the set safety limits. When the estimated values ​​of the cointegration coefficients at the current time at three or more scales exceed the safety limits, a ship-bridge collision alarm is triggered, and a final emergency alarm signal is pushed.

2. The ship bridge collision detection method according to claim 1, characterized in that, A low-pass filter is used to perform multi-level discrete wavelet transform on the displacement response data to decompose it into multi-scale approximate signals of each level of the displacement response. The specific formula is as follows: ； wherein is the first layer of the multi-scale approximation signal; is the coefficient of the low-pass filter for extracting the low-frequency component; n is the number of acquisition points of the displacement response signal; is the filter length, k = n .

3. The ship bridge collision detection method according to claim 2, characterized in that, The test model for stationarity testing of the L-level multi-scale approximation signal of the two sets of displacement response data is as follows: ； wherein is the original displacement response data, is a time index, is a position index of the lag order, is a constant term, is a trend coefficient, is a unit root coefficient, is a model coefficient, is defined as , is a white noise process, is a lag order of an autoregressive process; After being identified as a non-stationary sequence through stationarity testing, Johansen cointegration tests were performed on the non-stationary signals of the same scale on both sides, yielding multiple cointegration equations: ； ； ︙ ； wherein, and are the multi-scale approximate signals of the same height of each layer on both sides of the pier transversely, and are the approximate signals of the corresponding scale decomposition of the original displacement response data, and are the corresponding cointegration coefficients, is a random disturbance term.

4. The ship bridge collision detection method according to claim 3, characterized in that, The cointegration equation is taken as a state equation, and cointegration coefficients of the cointegration equation constitute an initial state vector x The multi-scale approximation signals of each layer are taken as observation variables, and the Kalman filtering algorithm is used for calculation to obtain the cointegration coefficient estimation value of each scale at the current time in real time. Specifically, ； ； wherein is a state vector of cointegration coefficients, is a state vector of cointegration coefficients, is a rate of change over time, is a cointegration coefficient; is a zero-mean stationary process noise with covariance Q, is an observation noise with covariance matrix R, and the observation matrix is ; The estimated values ​​of the cointegration coefficients at each scale are obtained in real time through prediction updates and measurement updates, as detailed below: Forecast Update: ； ； Measurement Update: ； ； ； wherein is the observation matrix at time k, is the transpose of the observation matrix at time k, is the state vector at time k, is the co-integration coefficient estimate at time k, is the co-integration coefficient forecast at time k, is , are , the error covariance matrix associated with, is the identity matrix, is the observation variable, is the Kalman gain matrix, is the error covariance matrix at time k, the process noise at time k, the observation noise at time k.

5. The ship bridge collision detection method of claim 1, wherein, A fixed safety limit is set using a statistical control process method; when the estimated value of the cointegration coefficient remains within the safety limit, no abnormal alarm is triggered, and when it exceeds the safety limit, an automatic alarm is triggered; when the alarm results of the four scale cointegration combinations are combined, a bridge collision alarm decision is made when three or more cointegration combination alarms are set.

6. The ship bridge collision detection method according to claim 5, characterized in that, The method for setting fixed safety limits using statistical control processes is specifically formulated as follows: ； wherein upper and lower limits of the safety margin, respectively, denotes the average of the estimated cointegration coefficients; is the value of the standard normal distribution corresponding to a 95% confidence level; denotes the standard deviation of the estimated cointegration coefficients; is the sample size.

7. A ship bridge collision detection system characterized by, include: The data acquisition module is used to collect the original displacement response data at the same height on both sides of the impacted bridge pier. The two sets of displacement response data are subjected to L-level discrete wavelet transform to decompose the two sets of displacement response data into multi-scale approximate signals of each level. The data processing module is used to perform stationarity tests on the L-level multi-scale approximate signals of two sets of displacement response data, and to perform Johansen cointegration tests on the selected non-stationary signals of the same scale on both sides to obtain L cointegration equations. Each cointegration equation is taken as a state equation, and the cointegration coefficients of the cointegration equation constitute an initial state vector x Each layer multi-scale approximation signal is taken as an observation variable, and the Kalman filtering algorithm is used for calculation to obtain the cointegration coefficient estimation value of each scale at the current time in real time. The alarm module is used to compare the estimated values ​​of the cointegration coefficients at the current time at each scale with the set safety limits. When the estimated values ​​of the cointegration coefficients at the current time at three or more scales exceed the safety limits, a ship-bridge collision alarm is triggered, and a final emergency alarm signal is pushed.

8. A computer device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the processor to perform the method of any one of claims 1-7. The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, When the computer program is loaded by the processor, it is able to perform the steps of the method according to any one of claims 1 to 6.

Images