Simulation-based ship to-be-locked bearing capacity evaluation method

Abstract

The invention provides a simulation-based ship lock waiting bearing capacity evaluation method, which comprises the following steps of: S1, collecting ship lock operation data, and preprocessing the collected data; s2, constructing a model by using an Agent-based modeling and simulation algorithm according to an actual structure and an operation rule of the ship lock; s3, calculating an average daily lockage ship parameter, a lockage demand, a to-be-locked ship number parameter and a ship lock trafficability parameter according to the collected data; s4, selecting historical ship lock operation data as a verification sample, and adjusting model parameters by comparing a simulation result in S2 with actual data in S3; and S5, simulating different working conditions according to the parameters of the simulation model adjusted in the step S4, analyzing the ship backlog condition, evaluating the to-be-locked bearing capacity of the ship lock, and after evaluation data of the to-be-locked bearing capacity is obtained according to an evaluation result, carrying out ship lock scheduling optimization in the aspects of navigation frequency, an overhaul plan, severe weather influence coping and the like. And the navigation efficiency of the ship lock is improved to the maximum extent.

Description

Technical Field

[0001] The present invention relates to the field of the waiting lock bearing capacity of ship locks, and particularly to an evaluation method for the waiting lock bearing capacity of ships based on simulation. Background Art

[0002] The passing capacity of the hub cannot meet the demand for ships to pass through the dam, resulting in a large number of ships waiting for the lock in the waters upstream and downstream of the hub. The contradiction between the passing capacity and the demand for passing through the dam will be further exacerbated, and there will be a phenomenon that a large number of ships wait for the lock for a long time in the navigation scheduling waters of the ship lock. According to the navigation scheduling principle of "first come, first served", ships arriving later need to wait for the same type of ships that arrived earlier to pass through the lock before they can pass through. Therefore, the increase in the number of waiting ships directly leads to a corresponding extension of the waiting time of the ships. With the development of inland waterway shipping, the frequency of use of ship locks is increasing, and the problem of ship waiting lock backlog is becoming increasingly serious. The existing evaluation methods for the waiting lock bearing capacity of ship locks mostly rely on experience or simplified models, which cannot accurately reflect the actual operation situation and lack adaptability to complex working conditions. Summary of the Invention

[0003] The main purpose of the present invention is to provide an evaluation method for the waiting lock bearing capacity of ships based on simulation. By establishing a simulation model, the waiting lock bearing capacity of the ship lock under different working conditions can be accurately evaluated, providing a scientific basis for the scheduling and management of the ship lock.

[0004] To solve the above technical problems, the technical solution adopted by the present invention is: an evaluation method for the waiting lock bearing capacity of ships based on simulation, the method comprising: S1. Collect the operation data of the ship lock and preprocess the collected data; S2. Construct a model by using the Agent-based modeling and simulation algorithm according to the actual structure and operation rules of the ship lock; S3. Calculate the parameters of the ships passing through the lock per day on average, the passing demand, the parameters of the number of waiting ships, and the passing capacity parameters of the ship lock according to the collected data; S4. Select the historical operation data of the ship lock as the verification sample, and adjust the model parameters by comparing the simulation results in S2 with the actual data in S3; S5. Carry out simulations under different working conditions according to the simulation model parameters adjusted in the above S4 step, analyze the ship backlog situation, evaluate the waiting lock bearing capacity of the ship lock, and reasonably adjust the navigation frequency of the ship lock according to the evaluation results; S6. Carry out simulations under different working conditions according to the simulation model parameters adjusted in the above S4 step. According to the number of waiting ships simulated under different working conditions, when the number of waiting ships exceeds the bearing capacity, take measures in time to evacuate the waiting ships.

[0005] In the preferred solution, the operation data of the ship lock includes the ship type, quantity, tonnage, and lock passing time.

[0006] In the preferred solution, the preprocessing step in step S1 includes: Using a clustering-based outlier detection algorithm to identify and remove error data and outliers. This algorithm measures the similarity between data points and determines data points far from the cluster as outliers; Performing normalization processing on the data using a quantile-based normalization algorithm, mapping data of different magnitudes to a unified interval to provide accurate data for subsequent modeling.

[0007] In the preferred solution, the simulation model building in step S2 includes: a ship generation module, a scheduling module, and a lock passage module; Ship generation module: With the help of a ship arrival prediction algorithm based on the HMM model, according to the characteristics of the ship arrival time series in historical data, predicting the ship arrival probability at different time periods, and then randomly generating ship arrival events, where HMM is the hidden Markov model; Scheduling module: Using an optimization scheduling algorithm based on the genetic algorithm to determine the lock passage priority; Lock passage module: Using a ship lock passage process modeling algorithm based on Petri nets to accurately describe the complex process of ships entering and leaving the lock and the resource occupancy situation.

[0008] In the preferred solution, the steps of building the simulation model are: Ship generation module: Cleaning the historical ship arrival time series data, dividing it at a certain time interval and counting the number of ships arriving within each interval; determining the number of states of the HMM model , initializing the state transition probability matrix , the observation probability matrix and the initial state probability vector ; training the HMM using the Baum - Welch algorithm and calculating the forward variable : ; ; the backward variable : ; ; the auxiliary variable : ; : ; and updating the model parameters , , , until the model parameters converge; use the trained HMM to predict the number of ships arriving at the next moment, and generate ship arrival events through a random number generator; Scheduling module: Use integer coding to encode the ship scheduling plan into an integer array with a length equal to the number of ships ; Randomly generate initial scheduling plans as the population; Define the fitness function : ; Calculate the fitness value of each scheduling plan, where , , are weight coefficients and , is the arrival time of the th ship, is the service time of the th ship, is the priority adjustment time of the th ship; Use the roulette wheel selection method, according to the selection probability : ; Select individuals, where is the fitness value of the th individual; Randomly select two individuals from the new population as parents, perform crossover operations using the partially matched crossover method, and then perform mutation operations on the individuals in the next generation population with a mutation probability using the swap mutation method; Repeat the fitness calculation, selection, crossover, and mutation operations until the termination conditions of the maximum number of iterations or the convergence of the population fitness value are met, and obtain the optimal ship scheduling order.

[0009] Lock passing module: Determine the places, transitions, and connection relationships of the Petri net according to the actual structure and operating rules of the lock, set the initial markings of the places; Define the transition triggering rules, and change the marking quantity of the places according to the rules; Construct a reachability graph for resource occupancy analysis; Define performance indicators, and calculate the average waiting time : ; where is the waiting time of the th ship in the waiting area, is the total number of ships passing through the lock within the statistical time period; Calculate the lock utilization rate : ; where is the total duration of the statistical time period, is the moment Usage status of the ship lock, when in use , when idle , to evaluate the efficiency of the ship lock passing process.

[0010] In the preferred solution, the steps of calculating the parameters of the ships passing through the lock per day, the passing demand and the number of ships waiting in the lock, and the passing capacity parameter of the ship lock in step S3 are as follows: Calculate the passing demand of ships per day: The number of ships passing through the lock throughout the year is M 2 ships, then the passing demand of ships per day is ship trips, is the passing cycle of the jth ship; the number of newly added ships waiting in the lock on the ith day is , the total number of ships waiting in the lock is , the passing capacity on the ith day is , after maintenance for i - 1 days, the passing cycle of the subsequent ships is forced to increase by days, and the passing demand of the ships passing through the two-dam ship lock on the ith day of maintenance is ship trips; Estimate the passing demand and the number of ships waiting in the lock: The passing demand on the th day , the number of newly added ships waiting in the lock on the th day , the number of ships waiting in the lock on the th day ; In the formula is the passing demand on the ith day, is the number of ships waiting in the lock on the ith day, is the number of newly added ships waiting in the lock on the ith day; Among them is the seasonal factor of the passing demand, is the passing capacity on the th day, is the number of ships waiting in the lock on the th day, and the number of ships waiting in the lock The iteration convergence condition is ; Calculate the passing capacity of the ship lock: For a multi-line ship lock, there are a total of lines, and the number of non-out-of-service ship locks is , the passing capacity of the ship lock on the th day , among which the daily operation gate times of the th non-out-of-service ship lock is , and the average number of ships passing through each gate time is , is the relevant correction coefficient.

[0011] In the preferred solution, the model verification step in step S4 is as follows: Use the similarity measurement method based on the dynamic time warping distance to compare the simulation results with the actual data; If the similarity does not meet the preset threshold, use the simulated annealing algorithm to adjust the model parameters and re-verify until the model accuracy reaches the standard; The simulated annealing algorithm searches for the global optimal solution in the solution space by simulating the physical annealing process to avoid falling into the local optimum.

[0012] In the preferred solution, the model verification step in step S4 is as follows: Data preparation and simulation operation: Screen the data of representative time periods covering various working conditions from the historical lock operation database as verification samples, input them into the constructed simulation model, run the model and record the results such as the ship lock passing time series and the daily number of ships waiting for lock series output by the simulation; Similarity measurement based on the dynamic time warping distance: For the simulation and actual ship lock passing time series 、 , construct a distance matrix , , initialize the cumulative distance matrix , , When , When , And When , the DTW distance ; Calculate the DTW distance of the simulation and actual waiting ship number series in the same way; Set the similarity threshold , if the calculated DTW distance is greater than , it is determined that the similarity does not meet the requirements; Model parameter adjustment based on the simulated annealing algorithm: Determine the model parameters to be adjusted; Among them, the mean value 、standard deviation of the ship arrival interval time, the shape parameter 、scale parameter of the lock passing time distribution; Define the objective function , where is the DTW distance between the simulation and actual ship lock passing times, is the DTW distance between the simulation and actual waiting ship numbers, 、 are weight coefficients and ; Set the initial temperature ​, cooling coefficient and termination temperature ; at the current temperature , randomly generate new model parameter values, and calculate the objective function values under the new and old parameters , , calculate the acceptance probability , if or the random number is less than , then accept the new parameter value, otherwise keep the current value; lower the temperature , when stop the iteration; re-run the simulation model with the adjusted parameters, calculate the DTW distance again and compare it with the threshold , if the requirement is not met, continue to adjust until the similarity between the simulation result and the actual data reaches the preset threshold.

[0013] In the preferred solution, the steps of analyzing the ship backlog situation and evaluating the lock's waiting capacity include iterative calculation of the number of ships waiting for the lock and the evaluation of the waiting capacity; Iterative calculation of the number of ships waiting for the lock: After giving the number of ships waiting for the lock and the data of the lockage demand, perform iterative calculation on the formula for estimating the number of ships waiting for the lock to estimate the number of ships waiting for the lock in a certain period; Evaluation of the waiting capacity of the lock: Set different working condition scenarios in the verified simulation model to simulate the influence of different degrees of maintenance duration, different frequencies and durations of bad weather on the lock; Adopt the method combining Latin hypercube sampling and Bayesian network, repeat the simulation of each working condition multiple times to obtain a large number of sample data; Latin hypercube sampling can reduce the number of samplings while ensuring the representativeness of the samples, and the Bayesian network is used to analyze the causal relationship between each working condition factor and the number of ship backlogs; By analyzing the change trend, peak value, and mean statistical characteristics of the number of ship backlogs in these data, evaluate the waiting capacity of the lock under different working conditions, and provide a decision-making basis for the lock scheduling and management.

[0014] In the preferred solution, the steps for evaluating the waiting capacity of the lock are: Iterative calculation of the number of ships waiting for the lock: Define the initial number of ships waiting for the lock , the annual number of ships passing through the lock , the seasonal factor of the lockage demand , the lockage cycle of each ship , the daily lockage times of each non-idle lock in the multi-line lock , the average number of ships passing through each lockage and the correction coefficient ; According to the formula calculate the The lockage demand for a day, where is the number of days increased in the later ship lockage cycle due to maintenance, are relevant parameters; according to the formula calculate the number of ships waiting for lockage newly added on the th day, where , , is the number of non - suspended ship locks; according to the formula Start iterative calculation from until the preset period is reached to obtain the number of ships waiting for lockage every day during this period; Evaluation of the waiting - lock bearing capacity: Determine the working condition factors affecting the waiting - lock bearing capacity of the ship lock,; Among them, the maintenance duration of the ship lock , the frequency of bad weather , the duration of bad weather ; Divide the value range of each factor into non - overlapping intervals, and the probability of each interval is , randomly draw sample values from each interval to form sample points; According to the ship lock operation knowledge and experience, determine the causal relationship structure between each working condition factor and the number of ship backlogs to construct a Bayesian network, and use historical data and sampled sample data to calculate the conditional probability distribution of each node in the Bayesian network by using the maximum likelihood estimation or Bayesian estimation method ; Input the sampled sample points into the verified simulation model in sequence. Run the simulation model multiple times for each sample point and record the number of ship backlogs; Calculate statistical features such as the change trend, peak value, and mean value for the number of ship backlogs data simulated each time. Use the Bayesian network to analyze the influence degree of each working condition factor on the number of ship backlogs, and comprehensively evaluate the waiting - lock bearing capacity of the ship lock under different working conditions to provide a decision - making basis for ship lock scheduling and management.

[0015] The present invention provides a method for evaluating the waiting - lock bearing capacity of ships based on simulation. The method for evaluating the waiting - lock bearing capacity of ships based on simulation realizes the accurate evaluation of the waiting - lock bearing capacity of the ship lock by constructing a simulation model and applying various algorithms, and has significant advantages in the operation management of the ship lock.

[0016] In the data processing and model construction stage, comprehensively collect the operation data of the ship lock and preprocess it, and construct a model using the Agent-based modeling and simulation algorithm. Among them, the ship generation module uses the ship arrival prediction algorithm based on the Hidden Markov Model (HMM) to accurately capture the characteristics of the ship arrival time series, randomly generate ship arrival events, and can be more in line with the actual situation compared with traditional methods; the scheduling module adopts the optimized scheduling algorithm based on the genetic algorithm, comprehensively considers various factors to determine the priority of passing through the lock, and iteratively searches for the optimal scheduling order, effectively improving the ship lock scheduling efficiency; the passing-through-lock module uses the ship passing-through-lock process modeling algorithm based on Petri nets to clearly depict the ship passing-through-lock process and resource occupancy, providing strong support for analyzing the operation status of the ship lock.

[0017] In the model verification and parameter adjustment link, select the historical ship lock operation data as the verification sample, and use the similarity measurement method based on the Dynamic Time Warping (DTW) distance to compare the simulation results with the actual data, which can effectively handle the problem of inconsistent time series lengths and accurately measure the similarity between the two. If the similarity does not meet the preset threshold, use the simulated annealing algorithm to adjust the model parameters to avoid falling into local optima, ensure the model accuracy, and make the model more accurately reflect the actual operation of the ship lock.

[0018] In terms of the evaluation of the waiting lock bearing capacity, the method of iteratively calculating the number of waiting ships, through iterative calculation using a formula considering various actual factors, can accurately estimate the number of waiting ships in a certain period. The method of combining Latin hypercube sampling and Bayesian network is used to simulate different working conditions. Latin hypercube sampling reduces the number of samplings while ensuring the representativeness of the samples, and the Bayesian network is used to analyze the causal relationship between various working condition factors and the number of ship backlogs. Through the analysis of statistical characteristics such as the change trend, peak value, and mean value of the number of ship backlogs in a large number of sample data, the waiting lock bearing capacity of the ship lock under different working conditions can be comprehensively evaluated, providing a scientific and reliable decision-making basis for the ship lock scheduling and management, helping to reasonably arrange the ship lock maintenance plan, cope with the impact of bad weather, improve the overall operation efficiency of the ship lock, relieve the problem of ship backlogs, and promote the efficient development of inland waterway shipping. Brief Description of the Drawings

[0019] The present invention will be further described below in conjunction with the drawings and embodiments: Figure 1 It is the main view structure diagram of the cleaning process of the present invention; Detailed Embodiments Embodiment 1 As Figure 1 shown, a simulation-based method for evaluating the waiting lock bearing capacity of ships, the method includes: S1. Collect the operation data of the ship lock and preprocess the collected data; S2. Based on the actual structure and operation rules of the ship lock, construct a model using the Agent-based modeling and simulation algorithm; S3. Calculate the parameters of the vessels passing through the lock per day on average, the lockage demand, the parameters of the vessels waiting for lockage, and the lock passing capacity parameters according to the collected data; S4. Select the historical ship lock operation data as the verification sample, and adjust the model parameters by comparing the simulation results in S2 with the actual data in S3; S5. According to the simulation model parameters adjusted in the above S4 step, simulate different working conditions, analyze the situation of vessel backlog, evaluate the lockage waiting capacity of the ship lock, and reasonably adjust the navigation frequency of the ship lock according to the evaluation results; S6. According to the simulation model parameters adjusted in the above S4 step, simulate different working conditions, and according to the number of vessels waiting for lockage obtained by simulation under different working conditions, when the number of vessels waiting for lockage exceeds the carrying capacity, take measures in time to evacuate the vessels waiting for lockage.

[0020] The method for evaluating the lockage waiting capacity based on simulation provided by the present invention has significant advantages in the operation management of ship locks.

[0021] Accurate evaluation and prediction: In the data processing and model construction stage, comprehensively collect the ship lock operation data and perform preprocessing, and construct a model using the Agent-based modeling and simulation algorithm. The vessel generation module uses the vessel arrival prediction algorithm based on the Hidden Markov Model (HMM) to accurately capture the characteristics of the vessel arrival time series, randomly generate vessel arrival events, which is more in line with the actual situation than traditional methods; the scheduling module adopts the optimized scheduling algorithm based on the genetic algorithm, determines the lockage priority considering multiple factors, and iteratively searches for the optimal scheduling order to improve the ship lock scheduling efficiency; the lockage module uses the ship lockage process modeling algorithm based on Petri nets to clearly depict the ship lockage process and resource occupancy, providing strong support for analyzing the operation status of the ship lock. Through the model constructed by these algorithms, combined with the model verification and parameter adjustment links, it can accurately evaluate the lockage waiting capacity of the ship lock and effectively predict the change trend of the carrying capacity under different working conditions.

[0022] Optimize navigation scheduling: In the model verification and parameter adjustment link, select the historical ship lock operation data as the verification sample, and use the similarity measurement method based on the Dynamic Time Warping (DTW) distance to compare the simulation results with the actual data, effectively handle the problem of inconsistent time series lengths, and accurately measure the similarity between the two. If the similarity does not meet the preset threshold, use the simulated annealing algorithm to adjust the model parameters to avoid falling into local optima, ensure the model accuracy, and make the model more accurately reflect the actual operation of the ship lock. Based on the accurate evaluation and prediction results, the navigation frequency of the ship lock can be reasonably adjusted to maximize the navigation efficiency, reduce the vessel waiting time and backlog quantity under the premise of ensuring the safe operation of the ship lock.

[0023] Scientific decision-making support: In the evaluation of the waiting capacity of the lock, the method of iteratively calculating the number of waiting ships can accurately estimate the number of waiting ships in a certain period through iterative calculation with a formula considering various actual factors. The method combining Latin hypercube sampling and Bayesian network is used to simulate different working conditions. Latin hypercube sampling reduces the number of samplings while ensuring the representativeness of the samples, and the Bayesian network is used to analyze the causal relationship between various working condition factors and the number of ship backlogs. By analyzing the statistical characteristics such as the change trend, peak value, and mean value of the number of ship backlogs in a large amount of sample data, the waiting capacity of the lock under different working conditions is comprehensively evaluated, providing a scientific and reliable decision-making basis for lock scheduling and management, helping to reasonably arrange the lock maintenance plan, cope with the impact of bad weather, and promote the efficient development of inland waterway shipping.

[0024] In the optimal solution, the lock operation data includes ship type, quantity, tonnage, and lock passage time.

[0025] In the optimal solution, the preprocessing steps in step S1 include: Using the outlier detection algorithm based on clustering to identify and remove incorrect data and outliers. This algorithm determines outliers by measuring the similarity between data points and identifying data points far from the cluster. Using the normalization algorithm based on quantiles to normalize the data, mapping data of different magnitudes to a unified interval to provide accurate data for subsequent modeling.

[0026] Embodiment 2 Further illustrated in combination with Embodiment 1, as Figure 1 shown in the structure, the simulation model building in step S2 includes: a ship generation module, a scheduling module, and a lock passage module; Ship generation module: With the help of the ship arrival prediction algorithm based on the HMM model, according to the characteristics of the ship arrival time series in historical data, predict the ship arrival probability at different time periods, and then randomly generate ship arrival events, where HMM is the hidden Markov model; Scheduling module: Using the optimization scheduling algorithm based on the genetic algorithm to determine the lock passage priority; Lock passage module: Using the ship lock passage process modeling algorithm based on Petri nets to accurately describe the complex process of ships entering and leaving the lock and the resource occupancy situation.

[0027] In the optimal solution, the steps of building the simulation model are: Ship generation module: Clean the historical ship arrival time series data, divide it at a certain time interval and count the number of ships arriving in each interval; determine the number of states of the HMM model , initialize the state transition probability matrix , observation probability matrix and the initial state probability vector ; Train the HMM using the Baum - Welch algorithm and calculate the forward variables : ; ; backward variables : ; ; auxiliary variables : ; : ; and update the model parameters accordingly 、 、 , until the model parameters converge; Use the trained HMM to predict the number of ships arriving at the next moment, and generate ship arrival events through a random number generator; Scheduling module: Encode the ship scheduling plan into an integer array of length equal to the number of ships using integer encoding; Randomly generate initial scheduling plans as the population; Define the fitness function : ; Calculate the fitness value of each scheduling plan, where 、 、 are weight coefficients and , is the arrival time of the th ship, is the service time of the th ship, is the priority adjustment time of the th ship; Adopt the roulette wheel selection method, according to the selection probability : ; Select individuals, where is the fitness value of the th individual; Randomly select two individuals from the new population as parents, perform crossover operations using the partially matched crossover method, and then mutate the individuals in the next - generation population with a mutation probability The mutation operation is performed using the swap mutation method; the fitness calculation, selection, crossover, and mutation operations are repeated until the termination condition of the maximum number of iterations or the convergence of the population fitness value is met, and the optimal ship scheduling sequence is obtained.

[0028] Lock passing module: Determine the places, transitions, and connection relationships of the Petri net according to the actual structure and operation rules of the lock, and set the initial markings of the places; define the transition triggering rules and change the number of markings of the places according to the rules; construct a reachability graph for resource occupancy analysis; define performance indicators and calculate the average waiting time : ; where is the waiting time of the th ship in the waiting area, is the total number of ships passing through the lock within the statistical time period; calculate the lock utilization rate : ; where is the total duration of the statistical time period, is the usage status of the lock at time , when in use , when idle , to evaluate the efficiency of the lock passing process.

[0029] Detailed algorithm for step S2: 1. Ship generation module Data preprocessing: Clean the historical ship arrival time series data to remove outliers and incorrect data. Divide the time series at a certain time interval (such as 1 hour) and count the number of ships arriving in each interval.

[0030] Model initialization: Determine the number of states of the hidden Markov model (HMM). Assume that the ship arrival states are divided into three states: "low flow", "medium flow", and "high flow", that is . Initialize the state transition probability matrix , represents the probability of transitioning from state to state , , for example: ; Initialize the observation probability matrix , represents the probability of observing the number of ship arrivals as in state , , is the possible number of ship arrivals, determined by statistical analysis of historical data. Among them, assuming in the "low traffic" state, the probability of observing 0 ships arriving is 0.8, and the probability of 1 ship arriving is 0.2, etc. Initialize the initial state probability vector , represents the probability of being in state at the initial moment, , assuming the probability of being in the "low traffic" state at the beginning is 0.6, the probability of being in the "medium traffic" state is 0.3, and the probability of being in the "high traffic" state is 0.1.

[0031] Model training: Use the Baum - Welch algorithm to train the HMM. This algorithm is an iterative algorithm based on the Expectation - Maximization (EM) algorithm, which maximizes the probability of the model generating the observed data by continuously updating the model parameters , and . In each iteration, first calculate the forward variable and the backward variable .

[0032] Forward variable: ; ; where is the number of ship arrivals observed at time , is the length of the time series. The forward variable is used to calculate the probability of being in state at time and observing the first data.

[0033] Backward variable: ; ; The backward variable is used to calculate the probability of being in state at time and observing the data from to .

[0034] Calculate the auxiliary variables and based on the forward variable and the backward variable: ; ; represents the probability of being in state at time , Indicates at the moment From state Transfer to state Probability. Update the model parameters through these auxiliary variables: ; ; ; Repeat the above training steps until the model parameters converge, that is, the parameter change is less than a certain preset threshold.

[0035] Ship arrival prediction and generation: Using the trained HMM, predict the state at the next moment according to the state and state transition probability at the current moment, and then predict the number of ships arriving at the next moment according to the observation probability matrix. Among them, the current moment is in the "medium flow" state. According to the state transition probability matrix, it is predicted that there is a high probability of transferring to the "high flow" state at the next moment. Then, combined with the observation probability matrix in the "high flow" state, it is predicted that 3 ships may arrive at the next moment. According to the prediction result, use a random number generator to generate ship arrival events. Assume that it is predicted that 3 ships will arrive at the next moment, and 3 ship arrival time points are generated within a certain time range through a uniform distribution random number generator.

[0036] The forward variable and backward variable calculation formulas in the Baum - Welch algorithm are used to calculate the probabilities of being in different states at different times, providing a basis for subsequent updating of model parameters. The auxiliary variables and The calculation formulas of are used to count the probabilities of different states and state transitions, and then update the state transition probability matrix , the observation probability matrix and the initial state probability vector , so that the model can more accurately reflect the characteristics of the ship arrival time series. Finally, through the predicted number of ship arrivals and the random number generator, the random generation of ship arrival events based on the HMM is realized.

[0037] 2. Scheduling module Algorithm steps: Encoding scheme design: Encode the ship scheduling scheme, using integer encoding. Assume there are a total of ships, and use an integer array with a length of to represent the scheduling order. The elements in the array are ship numbers, and each number appears only once in the array. Among them, there are 5 ships, and the encoding [3, 1, 4, 2, 5] means that the 3rd ship passes through the lock first, the 1st ship passes through the lock second, and so on.

[0038] Population initialization: Randomly generate initial scheduling schemes as the population, is the population size, such as . Each scheduling plan is an integer array that conforms to the coding rules.

[0039] Fitness calculation: Define a fitness function to evaluate the quality of each scheduling plan. Considering factors such as ship type, tonnage, arrival time, etc., construct a fitness function : ; Among them, is the arrival time of the th ship, is the service time of the th ship (related to ship type and tonnage, where large cargo ships have a long service time and small passenger ships have a short service time), is the priority adjustment time of the th ship (which can be determined according to factors such as the urgency of the goods transported by the ship), , , are weight coefficients, determined through the analysis of historical data and expert experience, and . Calculate the fitness value of each scheduling plan. The lower the fitness value, the better the scheduling plan.

[0040] Selection operation: Use the roulette wheel selection method to select individuals from the population. Calculate the selection probability of each individual: ; Among them, is the fitness value of the th individual. According to the selection probability, use a random number generator to select individuals. The lower the fitness value of an individual, the higher the probability of being selected. Put the selected individuals into a new population.

[0041] Crossover operation: Randomly select two individuals from the new population as parents and perform a crossover operation using the partially mapped crossover (PMX) method. Among them, select parent individual A = [1, 2, 3, 4, 5] and parent individual B = [5, 4, 3, 2, 1], randomly select two crossover points. Suppose the crossover points are 2 and 4, and exchange the parts between the two crossover points of the two parent individuals to obtain offspring individuals A' and B'. Then, by resolving conflicts (ensuring that each ship number appears only once in the offspring), obtain the final offspring individuals. Put the offspring individuals into the next generation population.

[0042] Mutation operation: Perform a mutation operation on the individuals in the next generation population using the swap mutation method. With a certain mutation probability (such as Select an individual, randomly select two positions in the individual, and swap the ship numbers at these two positions. For example, for the individual [1, 2, 3, 4, 5], if positions 2 and 4 are selected for mutation, the mutated individual will be [1, 4, 3, 2, 5].

[0043] Iterative optimization: Repeat the operations of fitness calculation, selection, crossover, and mutation until the termination condition is met. The termination condition can be the maximum number of iterations (e.g., 100 iterations) or the convergence of the fitness value of the population (e.g., the change in the optimal fitness value of the population for 10 consecutive generations is less than a certain threshold). The finally obtained optimal individual is the optimal ship scheduling order.

[0044] Fitness function Comprehensively considering factors such as the arrival time, service time, and priority adjustment time of the ships, and weighting different factors through weight coefficients, it can comprehensively evaluate the advantages and disadvantages of the scheduling scheme. Selection probability calculation formula Determine the probability of each individual being selected according to the fitness value, implement the selection operation based on fitness, so that a better scheduling scheme has a greater probability of being selected into the next generation population. These formulas cooperate with each other, and through the iterative optimization of the genetic algorithm, search for the optimal ship scheduling order to achieve efficient scheduling.

[0045] 3. Lock passage module Algorithm steps: Petri net model construction: Determine the elements of the Petri net according to the actual structure and operation rules of the lock. Define places, such as "ship waiting area", "lock idle", "ship passing through the lock", etc., which represent the positions or resource states of the ships in different states. Define transitions, such as "ship enters the lock", "ship leaves the lock", etc., which represent the state transition events. Determine the connection relationship between places and transitions, and draw the Petri net model. Set the initial tokens for each place, representing the number of ships or resource states in each state at the initial moment. The initial token of the "ship waiting area" place is 0, and the initial token of the "lock idle" place is 1, indicating that the lock is idle and there are no ships waiting at the initial time.

[0046] State transition rules: Define the triggering rules of transitions according to the lock operation logic. Among them, when there are ships in the "ship waiting area" (there is a token in the place) and the "lock is idle" (there is a token in the place), the transition "ship enters the lock" can be triggered. After the transition is triggered, change the token quantity of the place according to the preset rules. Assume that the transition "ship enters the lock" is triggered, the token quantity of the "ship waiting area" place is reduced by 1, the token quantity of the "lock idle" place is reduced by 1, and the token quantity of the "ship passing through the lock" place is increased by 1, indicating that a ship enters the lock from the waiting area and starts passing through the lock.

[0047] Resource Occupancy Analysis: The reachability analysis of Petri nets is used to study the occupancy of lock resources. The Reachability Graph is a directed graph that describes all possible states of a Petri net. By constructing the reachability graph, the occupancy status of lock resources (such as chamber space, equipment, etc.) at different times can be analyzed. Among them, from the reachability graph, it can be determined how many ships are using the lock resources at a certain time and how many ships are waiting to use the resources.

[0048] Performance Index Calculation: Performance indices are defined to evaluate the efficiency of the lock passage process. Among them, the average waiting time is calculated : ; Among them, is the waiting time of the th ship in the waiting area, is the total number of ships passing through the lock during the statistical time period. By simulating the operation of the Petri net model, the waiting time of each ship is recorded, and then the average waiting time is calculated. The lock utilization rate can also be calculated : ; Among them, is the total duration of the statistical time period, is the usage status of the lock at time (when in use , when idle ). By simulating the Petri net model in different time periods, the usage time of the lock is statistically calculated, and thus the lock utilization rate is calculated.

[0049] Average Waiting Time Formula is used to measure the average waiting duration of ships in the lock waiting area, reflecting the impact of lock scheduling on the waiting time of ships, and is one of the important indicators for evaluating the operation efficiency of locks. The lock utilization rate formula is used to calculate the usage ratio of the lock within a certain time period, which can visually show the utilization degree of lock resources and help analyze whether the lock resources are fully utilized or there is overcrowding. Through the Petri net model and these formulas, the complex process of ships entering and leaving the lock and the resource occupancy situation can be accurately described, providing a basis for optimizing the operation of the lock.

[0050] Example 3 Further illustrated in combination with Example 1, as Figure 1 shown in the structure, the steps of calculating the parameters of the ships passing through the lock per day, the passing demand, the parameters of the ships waiting at the lock, and the lock passing capacity parameters in step S3 are as follows: Calculating the demand for ships passing through the lock per day: The number of ships passing through the lock throughout the year is M 2If the number of ships is ship - times, is the lock - passing cycle of the j - th ship; the number of ships waiting for the lock on the i - th day is , and the total number of ships waiting for the lock is , the ship - passing capacity on the i - th day is . After maintenance for i - 1 days, the lock - passing cycle of the later - arriving ships is forced to increase by days, and the ship demand passing through the two - dam lock on the i - th day of maintenance is ship - times; Estimate the lock - passing demand and the number of ships waiting for the lock: The lock - passing demand on the th day is , the number of ships waiting for the lock newly added on the th day is , and the number of ships waiting for the lock on the th day is ; In the formula is the lock - passing demand on the i - th day, is the number of ships waiting for the lock on the i - th day, is the number of ships waiting for the lock newly added on the i - th day; Among them is the seasonal factor of lock - passing demand, is the ship - passing capacity on the th day, is the number of ships waiting for the lock on the th day. The iterative convergence condition for the number of ships waiting for the lock is ; ; Calculate the lock - passing capacity: For multi - line locks, there are a total of lines, and the number of non - suspended locks is . The lock - passing capacity on the th day is . Among them, the daily operation lock - times of the th non - suspended lock is , and the average number of ships passing through each lock - time is , is the relevant correction coefficient.

[0051] In the optimal solution, according to the calculated data above, the model verification step in step S4 is: Adopt the similarity measurement method based on the dynamic time warping distance to compare the simulation results with the actual data; If the similarity does not meet the preset threshold, use the simulated annealing algorithm to adjust the model parameters and re - verify until the model accuracy reaches the standard; The simulated annealing algorithm searches for the global optimal solution in the solution space by simulating the physical annealing process to avoid falling into the local optimum.

[0052] Let the number of ships passing through the lock throughout the year be M 2 ships, then the average daily demand for ships passing through the lock is (ship trips), being the lock-through cycle of the jth ship. Let the number of newly arrived ships waiting for the lock on the ith day be , the total number of ships waiting for the lock be , the ship-passing capacity on the ith day be , after maintenance for i - 1 days, the lock-through cycle of the later arriving ships is forced to increase by days, and the demand for ships passing through the two dams' locks on the ith day of maintenance is ship trips.

[0053] According to the principle of lock operation organization, the total number of ships waiting for the lock on the current day is equal to the sum of the number of ships waiting for the lock on the previous day and the difference between the lock-through demand and the passing capacity on the current day. After comprehensively considering factors such as the selected period (season), etc., the lock-through demand and the number of ships waiting for the lock on the ith day can be estimated by the following formula: (1) In the formula is the lock-through demand on the ith day, is the number of ships waiting for the lock on the ith day, is the number of newly arrived ships waiting for the lock on the ith day, is the ship-passing capacity on the ith day, M is the number of ships passing through the lock throughout the year, and g is the seasonal factor of lock-through demand.

[0054] It can be seen from formula (1) that the condition for the iteration convergence of the number of ships waiting for the lock is , that is, when the ship-passing capacity is equal to the lock-through demand, the number of ships waiting for the lock will stabilize around a certain value.

[0055] 3. After giving the initial conditions, perform iterative calculations on formula (1) to estimate the number of ships waiting for the lock in a certain period. There are a total of L lines of multi-line locks, and the number of non-suspended locks is L - 1. The daily number of lock operations of the kth non-suspended lock is , and the average number of ships passing through each lock operation is , then the ship-passing capacity of the lock on the ith day can be expressed as (2).

[0056] (2).

[0057] Example 4 Further illustrate in combination with Example 1. As shown in Figure 1 the structure, the model verification step in step S4 is as follows: Data Preparation and Simulation Run: Select data for representative time periods covering various operating conditions from the historical lock operation database as verification samples, input them into the constructed simulation model, run the model, and record results such as the ship lock passage time series and the daily number of ships waiting for lock passage output by the simulation; Similarity Measurement Based on Dynamic Time Warping Distance: For the simulated and actual ship lock passage time series 、 ,construct a distance matrix , ,initialize the cumulative distance matrix , , at time , at time , and at time ,the DTW distance ;calculate the DTW distance between the simulated and actual daily number of ships waiting for lock passage in the same way; set the similarity threshold ,if the calculated DTW distance is greater than ,then it is determined that the similarity requirement is not met; Model Parameter Adjustment Based on Simulated Annealing Algorithm: Determine the model parameters to be adjusted; Among them, the mean 、standard deviation of the ship arrival interval time, the shape parameter 、scale parameter of the lock passage time distribution; Define the objective function ,where is the DTW distance between the simulated and actual ship lock passage times, is the DTW distance between the simulated and actual daily number of ships waiting for lock passage, 、 are weight coefficients and ; Set the initial temperature 、cooling coefficient and termination temperature ;at the current temperature ,randomly generate new model parameter values, calculate the objective function values 、 under the new and old parameters, calculate the acceptance probability ,if or the random number is less than ,then accept the new parameter value, otherwise keep the current value; reduce the temperature ,when ​​​​​​​Stop the iteration; re-run the simulation model with the adjusted parameters, calculate the DTW distance again and compare it with the threshold If the requirement is not met, continue to adjust until the similarity between the simulation results and the actual data reaches the preset threshold.

[0058] Description of step S4: 1. Data preparation and simulation run Select representative time period data from the historical lock operation database as verification samples to ensure that various working conditions such as different seasons and different flow rates are covered. These data should include key information such as the ship passing time, the number of ships waiting in the lock per day, and the actual number of ships passing through the lock. Input the prepared historical data into the simulation model constructed in step S2, run the simulation model, simulate the operation of the lock during the corresponding historical time period, and record the results such as the ship passing time series and the number of ships waiting in the lock per day output by the simulation.

[0059] This step is mainly data processing and model operation, without the application of specific mathematical formulas. Its purpose is to provide simulation data for subsequent comparative analysis and compare with the actual historical data to evaluate the accuracy of the model.

[0060] 2. Similarity measurement based on dynamic time warping (DTW) distance For the ship passing time series obtained from the simulation and the ship passing time series in the actual historical data ( and may be different), construct a distance matrix , where , represents the absolute value of the difference between the rd time point in the simulation sequence and the th time point in the actual sequence. Initialize a cumulative distance matrix , , for , ; for , . For and , . The DTW distance . Calculate the DTW distance between the simulation waiting ship number sequence and the actual waiting ship number sequence in the same way. Set a similarity threshold (where $0.5$, which can be adjusted according to the actual situation), if the calculated DTW distance is greater than , it is considered that the similarity between the simulation results and the actual data does not meet the requirements, and the model parameters need to be adjusted.

[0061] Off - matrix Calculation formula Used to measure the degree of difference between data at corresponding time points in two time series. The cumulative distance matrix Calculation formula, through the idea of dynamic programming, gradually calculates the minimum cumulative distance from the starting point to the current point considering different path selections. The final DTW distance Can comprehensively reflect the overall similarity degree of two time series on the time axis. Even if their lengths are inconsistent, it can effectively measure and help judge the matching degree between the simulation results and the actual data.

[0062] 3. Model parameter adjustment based on simulated annealing algorithm Determine the model parameters to be adjusted, such as the mean of the ship arrival interval time and standard deviation , the shape parameter of the lock - passing time distribution and scale parameter (assuming the lock - passing time follows a Weibull distribution), etc. Define the objective function , based on the DTW distance , where is the DTW distance between the simulation and the actual ship lock - passing time, is the DTW distance between the simulation and the actual number of ships waiting for the lock, and are weight coefficients (where , , set according to the degree of attention to different indicators), and . Set the initial temperature (such as $100$), the cooling coefficient (such as $0.95$) and the termination temperature (such as ). At the current temperature , randomly generate a new set of model parameter values, calculate the objective function value of the simulation results and the actual data under the new parameters and the objective function value under the current parameters . Calculate the acceptance probability , if or the random number is less than , then accept the new model parameter values; otherwise, keep the current parameter values. Lower the temperature , stop the iteration when , and the model parameters obtained at this time are the adjusted parameters. Use the adjusted parameters to re - run the simulation model, calculate the DTW distance again, and compare it with the threshold Compare. If the requirements are still not met, continue to adjust the parameters using the simulated annealing algorithm until the similarity between the simulation results and the actual data reaches the preset threshold.

[0063] Objective function By using the weighted DTW distance, two key indicators, namely the ship lockage time and the number of ships waiting for lockage, are comprehensively considered, and the difference degree between the model and the actual data can be comprehensively evaluated. Acceptance probability formula , based on the Boltzmann distribution. At the initial stage of the algorithm, a higher temperature makes it possible to accept a new solution with a certain probability even if it is worse, avoiding being trapped in a local optimal solution; as the temperature decreases, the probability of accepting a worse solution gradually decreases, and the algorithm gradually converges to the global optimal solution or an approximate global optimal solution. By continuously adjusting the model parameters and evaluating the objective function value, the model can more accurately reflect the actual operation of the ship lock.

[0064] Embodiment 5 Further described in combination with Embodiment 1. As Figure 1 shown in the structure, the steps of analyzing the ship backlog situation and evaluating the lockage capacity of the ship lock waiting for lockage include iterative calculation of the number of ships waiting for lockage and the evaluation of the lockage capacity waiting for lockage; Iterative calculation of the number of ships waiting for lockage: After giving the number of ships waiting for lockage and the lockage demand data, perform iterative calculation on the formula for estimating the number of ships waiting for lockage to estimate the number of ships waiting for lockage in a certain period; Evaluation of the lockage capacity waiting for lockage: Set different working condition scenarios in the verified simulation model to simulate the influence of different degrees of maintenance duration, different frequencies and durations of bad weather of the ship lock; Adopt the method combining Latin hypercube sampling and Bayesian network, repeat the simulation of each working condition multiple times to obtain a large number of sample data; Latin hypercube sampling can reduce the number of samplings while ensuring the representativeness of the samples, and the Bayesian network is used to analyze the causal relationship between each working condition factor and the number of ship backlogs; By analyzing the change trend, peak value, and mean statistical characteristics of the number of ship backlogs in these data, evaluate the lockage capacity waiting for lockage of the ship lock under different working conditions, and provide a decision-making basis for the ship lock scheduling and management.

[0065] In the preferred solution, the steps for evaluating the lockage capacity waiting for lockage of the ship lock are: Iterative calculation of the number of ships waiting for lockage: Define the initial number of ships waiting for lockage , the annual number of ships passing through the lock , the seasonal factor of lockage demand , the lockage cycle of each ship , the daily number of lockages of each non-idle ship lock in the multi-line ship lock , the average number of ships passing through each lockage and the correction coefficient ; Calculate the lockage demand for the th day according to the formula, where is the number of days by which the lockage cycle of ships in the later stage is increased due to maintenance, and are relevant parameters; calculate the number of ships waiting for lockage newly added on the th day according to the formula, where Calculate the number of ships waiting for lockage newly added on the th day, where , , and is the number of non - suspended locks; start iterative calculation from until the preset period is reached to obtain the number of ships waiting for lockage every day during this period; Waiting - lock bearing capacity evaluation: Determine the working condition factors affecting the waiting - lock bearing capacity of the lock,; Among them, the lock maintenance duration , the frequency of bad weather , and the duration of bad weather ; Divide the value range of each factor into non - overlapping intervals, with the probability of each interval being , and randomly draw sample values from each interval to form sample points; According to the lock operation knowledge and experience, determine the causal relationship structure between each working condition factor and the number of ship backlogs to construct a Bayesian network, and use historical data and sampled sample data to calculate the conditional probability distribution of each node in the Bayesian network using the maximum likelihood estimation or Bayesian estimation method ; ; Input the sampled sample points into the verified simulation model in sequence, run the simulation model multiple times for each sample point and record the number of ship backlogs; Calculate the statistical characteristics such as the change trend, peak value, and mean value of the number of ship backlogs data for each simulation, use the Bayesian network to analyze the influence degree of each working condition factor on the number of ship backlogs, and comprehensively evaluate the waiting - lock bearing capacity of the lock under different working conditions to provide a decision - making basis for lock scheduling and management.

[0066] Description of Step S5 1. Iteratively calculate the number of ships waiting for lockage Practical steps of the algorithm: Define the initial conditions, including the initial number of ships waiting for lockage , the annual number of ships passing through the lock , and the seasonal factor of lockage demand determined according to historical data and actual situations , the lockage cycles of each ship ( ), etc. At the same time, determine the parameters required to calculate the lock passing capacity, such as the daily operation times of each non - suspended lock in a multi - line lock , the number of ships passing through per lockage ( ), and the possible correction factor .

[0067] Based on the formula in the document, the lockage demand on the th day is calculated as follows: ; where represents the number of days by which the lockage cycle of ships in the later stage is increased due to maintenance (if there is no maintenance, this item is 0), is a relevant parameter (related to the calculation parameters of the lockage capacity). This formula comprehensively considers the ship lockage cycle, seasonal factors, and possible maintenance impacts, and calculates the ship lockage demand for the day.

[0068] The number of ships waiting for lockage newly added on the th day is calculated as follows: ; where is the ship passage capacity on the th day, ( , representing the number of non-idled locks). This formula obtains the number of ships waiting for lockage newly added on the day by taking the difference between the lockage demand and the ship passage capacity.

[0069] The number of ships waiting for lockage on the th day is calculated as follows: ; Starting from , iterative calculations are performed using the above formula. In each iteration, the number of ships waiting for lockage obtained from the previous iteration and the newly calculated number of ships waiting for lockage newly added are used to calculate the current number of ships waiting for lockage . The iterative calculations continue until a preset period (such as a month, a quarter, etc., assumed to be days) is reached, and the number of ships waiting for lockage for each day within this period is obtained.

[0070] The above three formulas are interrelated and jointly achieve the iterative calculation of the number of ships waiting for lockage. The calculation formula of comprehensively considers various factors affecting the lockage demand and provides a basis for subsequent calculations; The formula accurately estimates the number of vessels waiting for lockage each day within a certain period by iteratively accumulating the newly added number of vessels waiting for lockage, providing data support for subsequent analysis of vessel backlog situations.

[0071] 2. Evaluation of Lockage Carrying Capacity Latin hypercube sampling: Determine the operating condition factors affecting the lockage carrying capacity of the lock, such as the duration of lock maintenance , frequency of bad weather , and duration of bad weather , etc. Assume that the value range of each factor is known. Divide the value range of each factor into non - overlapping intervals (such as ), and the probability of each interval is . For each factor, randomly select a sample value from each interval to form a sample point. Among them, for the duration of lock maintenance , its value range is $[0, 30]$ days, divided into 10 intervals, each interval being 3 days. Randomly select a value from each interval, such as , , , etc. In this way, sample points are obtained, and each sample point contains the value combinations of all operating condition factors, which can greatly reduce the number of sampling times while ensuring the representativeness of the samples.

[0072] Construct a Bayesian network: Based on the domain knowledge and practical experience of lock operation, determine the causal relationship structure between each operating condition factor (such as , , ) and the number of vessels in backlog to construct the Bayesian network structure. Use historical data and the sample data obtained through Latin hypercube sampling, and use the maximum likelihood estimation or Bayesian estimation method to calculate the conditional probability distribution (CPD) of each node in the Bayesian network. Among them, for the node (number of vessels in backlog), its conditional probability distribution represents the probability distribution of the number of vessels in backlog given the duration of lock maintenance, frequency of bad weather, and duration of bad weather.

[0073] Simulation and data analysis: The sample points obtained by Latin hypercube sampling are input into the verified simulation model in sequence to simulate the operation of the lock under different working conditions. For each sample point, the simulation model is run multiple times (such as 50 times), and the number of ship backlogs obtained from each simulation is recorded. For the ship backlog data obtained from each simulation, its statistical characteristics such as change trend, peak value, and mean are calculated. The constructed Bayesian network is used to analyze the influence of various working conditions on the number of ship backlogs. Among them, Bayesian network reasoning is used to calculate the probability and amplitude of the increase in the number of ship backlogs when the lock maintenance time increases by a certain number of days; or when the frequency and duration of severe weather changes, the change in the number of ship backlogs. Based on the above analysis results, the waiting capacity of the lock under different working conditions is comprehensively evaluated. Among them, when the lock maintenance time is The frequency of severe weather is , duration of bad weather When the ship lock is in operation, the statistical characteristics of the simulation data and the results of the Bayesian network analysis are used to determine whether the waiting capacity of the ship lock is strong, medium or weak, and corresponding quantitative indicators (such as the average number of ships waiting for the lock, the maximum number of ships waiting for the lock, etc.) are given to provide a decision-making basis for the scheduling and management of the ship lock.

[0074] In Latin hypercube sampling, by reasonably dividing the interval and random sampling, the representativeness and diversity of the sample are guaranteed, the number of samplings is reduced, and the simulation efficiency is improved. In Bayesian network construction, the conditional probability distribution is calculated. The formula is used to describe the probability relationship between various operating factors and the number of ship backlogs. This formula can quantify the impact of different operating factors on the number of ship backlogs, providing a powerful tool for analyzing the waiting capacity of locks. Combining Latin hypercube sampling and Bayesian network methods, through simulation and data analysis, it can more comprehensively and accurately evaluate the waiting capacity of locks under different operating conditions, providing a scientific decision-making basis for lock scheduling and management.

[0075] After obtaining the evaluation data of the waiting lock carrying capacity, the ship lock scheduling can be optimized from the aspects of navigation frequency, maintenance plan, and response to the impact of severe weather, as follows: 1. Adjust the navigation frequency: Dynamically adjust according to the number of ships waiting for the lock and the demand for passing the lock. If the number of ships waiting for the lock continues to increase, the demand for passing the lock is large, and the lock's capacity allows, increase the navigation frequency to reduce the waiting time of ships; if the number of ships waiting for the lock is small and the demand for passing the lock is low, appropriately reduce the frequency to save resources. For example, when there are more ships in the peak tourist season, increase the navigation frequency; and reduce it in the off-season.

[0076] 2. Arrange maintenance plan: Arrange according to the data of the waiting lock's carrying capacity under different working conditions. If the carrying capacity is strong, a longer maintenance can be arranged during the period of low demand for lock passage; if the carrying capacity is sensitive to the maintenance time, a short-term segmented maintenance can be adopted. For example, in a specific off-season each year, a comprehensive maintenance can be arranged when the carrying capacity is high; for locks that are sensitive to carrying capacity, phased maintenance can be carried out during the off-peak period.

[0077] 3. Dealing with severe weather: Use the relationship between severe weather-related factors and the number of backlogged ships to provide early warning and dispatch. Before severe weather arrives, increase the frequency of navigation and speed up the passage of ships; make adjustments based on actual conditions during the period to ensure safety; evaluate facilities afterwards, resume navigation, and quickly evacuate backlogged ships. For example, speed up the passage of ships before a typhoon arrives, make adjustments based on safety conditions during the typhoon, and quickly resume and dispatch backlogged ships afterwards.

[0078] 4. Optimize the scheduling strategy: Based on the evaluation data, combined with the ship type, tonnage and urgency of the cargo, optimize the scheduling algorithm to determine the priority of passing the lock. For example, give priority to the transportation of emergency materials and large ships to pass the lock to improve the overall efficiency. At the same time, continuously monitor the number of ships waiting for the lock and the changes in the demand for passing the lock, and dynamically adjust the scheduling strategy.

[0079] 5. Coordinated dispatch of multiple locks: For multiple locks, coordinated dispatch is carried out according to the evaluation data of the waiting capacity of each lock. The ship flow of each lock is balanced to avoid congestion in some locks and idleness in others. The efficiency of the entire inland waterway shipping system is improved through information sharing and unified dispatch.

[0080] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A simulation-based method for evaluating the ship's waiting lock carrying capacity, characterized by: The method includes: S1. Collect the lock operation data and pre-process the collected data; S2. Based on the actual structure and operation rules of the ship lock, the model is constructed using the agent-based modeling and simulation algorithm; S3. Calculate the average daily ship parameters passing through the lock, the lock passing demand, the number of ships waiting for the lock, and the lock passing capacity parameters based on the collected data; S4, select historical ship lock operation data as verification samples, and adjust model parameters by comparing the simulation results in S2 with the actual data in S3; S5. Simulate different working conditions according to the simulation model parameters adjusted in the above step S4, analyze the waiting conditions of ships in the lock, evaluate the waiting capacity of the lock, and reasonably adjust the navigation frequency of the lock according to the evaluation results; S6. Simulate different working conditions according to the simulation model parameters adjusted in step S4 above. According to the number of ships waiting to be locked obtained by simulation under different working conditions, when the number of ships waiting to be locked exceeds the carrying capacity, take timely measures to evacuate the ships waiting to be locked.

2. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 1 is characterized by: Lock operation data include ship type, quantity, tonnage, and lock passage time.

3. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 1 is characterized by: The preprocessing steps in step S1 include: Use clustering-based outlier detection algorithms to identify and remove erroneous data and outliers. This algorithm measures the similarity between data points and determines data points far away from the cluster as outliers. A quantile-based normalization algorithm is used to normalize the data, mapping data of different magnitudes to a unified interval to provide accurate data for subsequent modeling.

4. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 1 is characterized by: The simulation model in step S2 includes: a ship generation module, a scheduling module and a lock passing module; Ship generation module: With the help of the ship arrival prediction algorithm based on the HMM model, according to the characteristics of the ship arrival time series in the historical data, the probability of ship arrival in different time periods is predicted, and then the ship arrival events are randomly generated, where HMM is a hidden Markov model; Scheduling module: uses an optimized scheduling algorithm based on genetic algorithm to determine the priority of gate-passing; Lock-passing module: The ship-passing-locking process modeling algorithm based on Petri net is used to accurately describe the complex process and resource usage of ships entering and exiting the lock.

5. According to claim 4, a method for evaluating the ship's waiting lock carrying capacity based on simulation is characterized in that: The model building steps are: Ship generation module: cleans the historical ship arrival time series data, divides it into certain time intervals and counts the number of ship arrivals in each interval; determines the state number of the HMM model , initialize the state transition probability matrix , observation probability matrix and the initial state probability vector ; Use the Baum-Welch algorithm to train the HMM and calculate the forward variables : ； ； Backward Variable : ； ； Auxiliary variables : ； ： ； And update the model parameters accordingly , , , until the model parameters converge; use the trained HMM to predict the number of ships arriving at the next moment, and generate ship arrival events through a random number generator; Scheduling module: Use integer coding to encode the ship scheduling plan into a length equal to the number of ships. An integer array of ; randomly generated Initial scheduling schemes are used as populations; the fitness function is defined : ； Calculate the fitness value of each scheduling scheme, where , , is the weight coefficient and , For the The arrival time of the ship, For the The service time of the ship, For the The priority adjustment time of the ship is set by using the roulette wheel selection method, according to the selection probability : ； Select individuals where For the The fitness value of each individual; randomly select two individuals from the new population as parents, use the partial matching crossover method to perform the crossover operation, and then perform the crossover operation on the individuals in the next generation population with the mutation probability. The mutation operation is performed by using the exchange mutation method; the fitness calculation, selection, crossover and mutation operations are repeated until the termination condition of the maximum number of iterations or the convergence of the population fitness value is met, and the optimal ship scheduling order is obtained; Lock-passing module: Determine the location, transition and connection relationship of the Petri net according to the actual structure and operation rules of the lock, and set the initial mark of the location; Define the change triggering rules and change the number of place marks according to the rules; Build a reachability graph to analyze resource usage; define performance indicators and calculate average waiting time : ； in For the The waiting time of the ship in the waiting area, The total number of ships passing through the lock during the statistical period; Calculate the lock utilization rate : ； in is the total duration of the statistical time period, For the moment The use status of the lock, when in use , when you are free , to evaluate the efficiency of the lock passage process.

6. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 1 is characterized by: The steps of calculating the average daily ship parameters passing through the lock, the lock passing demand and the number of ships waiting for the lock, and the lock passing capacity parameters in step S3 are as follows: Calculate the average daily demand for ships passing through the lock: The number of ships passing through the lock all year round is M2, so the average daily demand for ships passing through the lock is Ship times, is the lock-passing cycle of the jth ship; the number of new ships waiting for the lock on the i-th day is The total number of ships waiting for lock is , the ship passing capacity on the i-th day is After the maintenance i-1 day, the ship's lock-passing cycle was forced to increase. The demand for ships passing through the Liangba ship locks on the i-th day of maintenance is Ship number; Estimation of the demand for locks and the number of ships waiting for locks: Daily gate demand , No. The number of new ships waiting for lock per day , No. Number of ships waiting for lock per day ; In the formula is the gate-passing demand on the i-th day, is the number of ships waiting for lock on the i-th day, is the number of new ships waiting for lock on the i-th day; in is the seasonal factor of gate demand, For the The ship's ability to pass through the sky, For the Number of ships waiting for lock per day, Number of ships waiting for lock The iterative convergence condition is ; Calculate the lock capacity: For multi-line locks, the total Line, the number of locks not closed is , No. Lock capacity per day , among which The number of daily lock operations of the non-suspended ship locks is The average number of ships passing through each lock is , is the correlation correction coefficient.

7. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 6 is characterized by: The model verification step in step S4 according to the above calculated data is: A similarity measurement method based on dynamic time warping distance is used to compare simulation results with actual data; If the similarity does not meet the preset threshold, the simulated annealing algorithm is used to adjust the model parameters and re-verify until the model accuracy meets the standard; The simulated annealing algorithm simulates the physical annealing process to find the global optimal solution in the solution space and avoid falling into the local optimum.

8. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 7 is characterized by: The model validation steps in step S4 are: Data preparation and simulation operation: Select representative time period data covering various working conditions from the historical ship lock operation database as verification samples, input them into the constructed simulation model, run the model and record the simulation output results such as the ship passing time series and the daily number of ships waiting for the lock; Similarity metric based on dynamic time warping distance: for simulated and actual ship passing time series , , construct the distance matrix , , initialize the cumulative distance matrix , , hour , hour , and hour , DTW distance ; Use the same method to calculate the DTW distance between the simulation and the actual number of ships waiting for the lock; set the similarity threshold , if the calculated DTW distance is greater than , then it is determined that the similarity does not meet the requirements; Model parameter adjustment based on simulated annealing algorithm: determine the model parameters that need to be adjusted; The mean of the ship arrival intervals is , Standard Deviation , shape parameter of gate-passing time distribution , scale parameters ; Define the objective function ,in is the DTW distance between the simulated and actual ship passing time, is the DTW distance between the simulated and actual number of ships waiting for lock, , is the weight coefficient and ; Set initial temperature , Temperature reduction coefficient and termination temperature ; At the current temperature Under this condition, randomly generate new model parameter values ​​and calculate the objective function values ​​under the new and old parameters , , calculate the acceptance probability ,like or random number Less than , then accept the new parameter value, otherwise keep the current value; reduce the temperature ,when Stop the iteration when ; rerun the simulation model with the adjusted parameters, calculate the DTW distance again and compare it with the threshold If the requirements are not met, the simulation results will be adjusted until the similarity between the simulation results and the actual data reaches the preset threshold.

9. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 8, characterized in that: The steps of analyzing the backlog of ships and evaluating the waiting capacity of the locks include iterative calculation of the number of waiting ships and evaluation of the waiting capacity; Iterative calculation of the number of ships waiting for the lock: after the number of ships waiting for the lock and the lock-passing demand data are given, the formula for estimating the number of ships waiting for the lock in claim 6 is iteratively calculated to estimate the number of ships waiting for the lock at a certain time; Evaluation of the lock-waiting carrying capacity: Different working conditions are set in the verified simulation model to simulate the maintenance time of the locks of different degrees and the impact of severe weather of different frequencies and durations; The Latin hypercube sampling method combined with the Bayesian network is used to repeatedly simulate each working condition and obtain a large amount of sample data; Latin hypercube sampling can reduce the number of samplings while ensuring sample representativeness, and Bayesian networks are used to analyze the causal relationship between various operating factors and the number of ship backlogs; By analyzing the changing trends, peak values, and mean statistical characteristics of the number of ship backlogs in these data, the waiting capacity of the locks under different working conditions is evaluated, providing a decision-making basis for lock scheduling and management.

10. The method for evaluating the ship's waiting lock carrying capacity based on simulation according to claim 9, characterized in that: The steps to assess the lock's waiting capacity are: Iterative calculation of the number of ships waiting for lock: clarify the initial number of ships waiting for lock , Number of ships passing through the locks throughout the year , seasonal factors of gate demand , each ship lock cycle , Daily operation times of each non-stop ship lock in the multi-line ship lock , Average number of ships passing through each lock And the correction factor ; According to the formula in claim 6 Calculate the The daily gate crossing demand, including The maintenance will increase the number of days of ship lock passage. is the relevant parameter; according to the formula Calculate the The number of new ships waiting for locks increased by , , is the number of locks that have not been closed; according to the formula from Start iterative calculation until the preset period is reached, and obtain the number of ships waiting for lock every day in the period; Evaluation of the waiting lock bearing capacity: determine the working conditions that affect the waiting lock bearing capacity; Among them, the maintenance time of the ship lock Frequency of severe weather , duration of bad weather ; Divide the value range of each factor into There are non-overlapping intervals, and the probability of each interval is , randomly draw sample values ​​from each interval Sample points; Determine the factors of various operating conditions and the number of ships in backlog based on the knowledge and experience of lock operation The causal relationship structure of the Bayesian network is used to construct the Bayesian network, and the conditional probability distribution of each node in the Bayesian network is calculated by using the historical data and sample data, and the maximum likelihood estimation or Bayesian estimation method is used. ; The sample points obtained by sampling are input into the verified simulation model in sequence, and the simulation model is run multiple times for each sample point and the number of ship backlogs is recorded; The changing trend, peak value and mean statistical characteristics of the ship backlog data of each simulation are calculated, and the Bayesian network is used to analyze the influence of various operating factors on the ship backlog. The waiting capacity of the lock under different operating conditions is comprehensively evaluated to provide a decision-making basis for lock scheduling and management.

Images