Wharf ship operation forecasting and early warning method

Abstract

The invention discloses a wharf ship operation forecasting and early warning method, and relates to the technical field of early warning, input data is classified according to data types, a correlation coefficient, a weight and an abnormal coefficient are calculated based on each type of input data, a risk score is finally obtained, weighted summation is carried out on the risk score of each type of input data, and the risk score of each type of input data is calculated. And a comprehensive risk score of a future wharf ship operation plan is obtained, so that accurate early warning is realized. Mutual influence among different input data can be comprehensively considered, weights of various types of data are dynamically adjusted, important factors are ensured to get due attention in prediction and risk assessment, abnormal conditions of the input data can be monitored in real time, and preventive measures are automatically generated. Compared with the prior art, the prediction accuracy and the risk assessment comprehensiveness are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of early warning technology, and in particular to a forecast and early warning method for ship operations at a dock. Background Art

[0002] Ship operations at the terminal are a key link in the modern port logistics system, involving a series of operations such as berthing, loading and unloading of cargo, and leaving the port. With the rapid development of global trade, the throughput of the terminal continues to increase, and the complexity of operations has also increased. In order to ensure the safety, efficiency and orderliness of ship operations at the terminal, it is necessary to accurately predict future operations, identify potential risk factors in a timely manner, and take corresponding preventive measures.

[0003] Existing methods for forecasting ship operations at ports usually rely on inputting multiple parameter data (such as meteorological conditions, tide height, wind speed, number of ship arrivals, etc.) to predict operation plans for a period of time in the future. However, these methods have the following significant limitations: Although existing prediction models can make predictions based on multiple parameter data, they do not fully consider the mutual influence between different input data and their effects on the prediction results. For example, some input data may have a greater impact on the prediction results, while other data may have a relatively small impact. This prediction method without weight distribution may lead to inaccurate prediction results and fail to fully reflect the actual situation.

[0004] Existing risk assessment methods usually only focus on a single factor (such as weather, tide, etc.), ignoring the interaction between different input data and failing to fully identify potential risks. This makes it easy for the system to miss important risk factors when facing a complex and changing environment.

[0005] Therefore, the prior art is in urgent need of a technical solution for a method for forecasting and warning terminal ship operations. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a method for forecasting and warning of ship operations at a terminal, which specifically comprises the following steps: Step S1, collecting terminal ship data through at least two sources, and performing preprocessing on the data; Step S2: using the preprocessed data as input data and using the pre-built forecasting model to forecast the ship operation situation at the terminal in the future, and formulating a future ship operation plan at the terminal according to the forecasting results of the forecasting model; Step S3, analyzing the risk level of the future terminal ship operation plan to obtain a comprehensive risk score of the future terminal ship operation plan; Step S3a, calling the input data of the current forecast model, classifying the input data according to the data type to obtain each type of input data, and arbitrarily sorting each type of input data to obtain the first to nth types of input data; Step S3b, analyzing the correlation between each type of input data to obtain the correlation coefficient between each type of input data; Step S3c, assigning a weight to each type of input data according to the correlation coefficient of each type of input data; Step S3d, analyzing the abnormality degree of each type of input data to obtain the abnormality coefficient of each type of input data; Step S3d1, based on the statistical analysis of historical data, a threshold interval of the actual observation value of each data in each type of input data is obtained, wherein the threshold interval includes an upper threshold limit and a lower threshold limit, and based on the upper threshold limit and the lower threshold limit, an intermediate threshold value is calculated; Step S3d2, using the threshold interval of the actual observation value of each data in each type of input data to determine the actual observation value of each data in each type of input data, if the actual observation value of the current data in each type of input data is within the threshold interval, the actual observation value of the current data is classified as a normal value; if the actual observation value of the current data in each type of input data is not within the threshold interval, the actual observation value of the current data is classified as an abnormal value; Step S3d3, counting the number of outliers in each type of input data, and calculating the degree of deviation of each outlier from the intermediate threshold, to obtain the deviation value of each outlier; Step S3d4, summing the deviation values ​​of each abnormal value in each type of input data to obtain the abnormal coefficient of each type of input data; Step S3e, comprehensively analyzing the correlation coefficient between each type of input data, the weight of each type of input data, and the abnormal coefficient of each type of input data, and calculating the risk score of each type of input data; Step S3f, summing the risk scores of each type of input data to obtain a comprehensive risk score of the future terminal ship operation plan; Step S4: set a warning threshold for the comprehensive risk score. If the current comprehensive risk score is greater than or equal to the warning threshold, an alarm is issued, and preventive measures for future terminal ship operation plans are formulated based on the current comprehensive risk score; if the current comprehensive risk score is less than the warning threshold, no alarm is required.

[0007] Furthermore, the degree of deviation of each outlier from the intermediate threshold is calculated to obtain the deviation value of each outlier, which specifically includes: Step S3d31, obtaining historical data of each type of input data, and calculating the standard deviation based on the historical data; Step S3d32, calculating the historical mean based on the historical data; Step S3d33, calculate the deviation value of each outlier according to the standard deviation, mean, actual observed value of the outlier, intermediate threshold, upper threshold, and lower threshold.

[0008] Furthermore, the calculation formula of the deviation value is: ; Where D i represents the deviation value of the i-th type of input data, x (i,j) represents the actual observed value of the jth outlier in the i-th input data, T m represents the middle threshold, U represents the upper threshold, and L represents the lower threshold. represents the weight coefficient, n represents the total number of categories of input data, and m represents the total number of outliers. represents the historical mean, Represents standard deviation.

[0009] Furthermore, the correlation between each type of input data is analyzed to obtain the correlation coefficient between each type of input data, specifically including: Step S3b1, obtaining the actual observation value of each data in each type of input data, normalizing the actual observation value of each data and taking the average, to obtain the comprehensive observation value of each type of input data; Step S3b2, repeating step S3b1 to obtain at least two groups of comprehensive observation values ​​of each type of input data, and constructing a data set of each type of input data based on the at least two groups of comprehensive observation values ​​of each type of input data; Step S3b3: Based on the data set of each type of input data, calculate the correlation coefficient between each type of input data.

[0010] Furthermore, each type of input data is weighted according to its correlation coefficient, specifically including: Step S3c1, summing the correlation coefficients between the first category of input data and the input data of all other categories to obtain a first sum of correlation coefficients; and so on, until the correlation coefficients between the nth category of input data and the input data of all other categories are summed to obtain an nth sum of correlation coefficients; Step S3c2, summing up all the correlation coefficients calculated in step S3c1 to obtain a total correlation coefficient; Step S3c3: assign a weight to each type of input data according to the ratio of the sum of each correlation coefficient to the total correlation coefficient.

[0011] Furthermore, the calculation formula for the weight of each type of input data is: ; In the formula, represents the weight of the i-th type of input data, S i represents the sum of the correlation coefficients of the i-th type of input data, S t represents the overall correlation coefficient.

[0012] Furthermore, the risk score calculation formula for each type of input data is: ; In the formula, R i represents the risk score of the i-th type of input data, represents the weight of the i-th type of input data, S i represents the sum of the correlation coefficients of the i-th type of input data, Y i represents the abnormal coefficient of the i-th type of input data, Represents the adjustment factor.

[0013] Furthermore, the calculation formula of the comprehensive risk score is: ; In the formula, R s represents the comprehensive risk score, R i represents the risk score of the i-th category of input data, and n represents the total number of categories of input data.

[0014] The embodiments of the present invention have the following technical effects: The present invention categorizes the input data according to the data type, and calculates the correlation coefficient, weight and abnormal coefficient based on each type of input data, and finally obtains the risk score, and then weighted sums the risk score of each type of input data to obtain the comprehensive risk score of the future terminal ship operation plan, thereby achieving accurate early warning. This method can not only comprehensively consider the mutual influence between different input data, dynamically adjust the weights of various types of data, and ensure that important factors are given due attention in prediction and risk assessment, but also monitor the abnormal situation of input data in real time, discover potential risks in advance, and automatically generate preventive measures. Compared with the prior art, the present invention significantly improves the accuracy of prediction and the comprehensiveness of risk assessment, enhances the adaptability and robustness of the system, and is particularly suitable for complex and changeable port environments, helping managers to respond to risks in a timely manner and ensure the safety and efficient operation of terminal ship operations.

[0015] By calculating the anomaly coefficient based on the intermediate threshold and the upper and lower limits, this method can effectively classify the data into normal and abnormal categories, and more accurately evaluate the degree of abnormality of each type of input data. Specifically, this method not only takes into account the degree of deviation of the actual observation value from the intermediate threshold, but also emphasizes that data close to the upper and lower limits are more likely to become anomalies, while data close to the intermediate threshold is more likely to be considered normal values. It can more accurately identify abnormal data, improve the accuracy of anomaly detection, enhance the comprehensiveness of risk assessment, and reduce the false alarm rate, thereby providing more reliable and effective early warning support for terminal ship operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 The present invention provides a flowchart of a method for forecasting and warning terminal ship operations. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0019] Embodiment 1: Figure 1 As shown, the present invention provides a method for forecasting and warning of dock ship operations, comprising the following steps: Step S1: collecting terminal ship data through at least two sources and performing preprocessing on the data.

[0020] In this embodiment, the terminal vessel data is collected from the following three main sources: Weather station: provides real-time weather data, including wind speed, wind direction, temperature, humidity, air pressure, etc. Weather data is crucial to the safety and efficiency of ship operations at the terminal, especially in severe weather conditions such as strong winds, heavy rain or dense fog, which may affect the berthing and loading and unloading operations of ships; Tidal observatory: provides data on tide height, tidal change rate, etc. Tidal data directly affects the berthing and departure time of ships, especially in ports with large tidal changes. Reasonable tidal planning can avoid accidents such as ship grounding or collision; Automatic Identification System (AIS): Provides dynamic information of ships entering and leaving the port, including ship name, ship type, speed, course, estimated time of arrival (ETA), estimated time of departure (ETD), etc. AIS data helps managers understand the real-time dynamics of ships in the port and arrange berths and operation plans reasonably.

[0021] Preprocessing includes data cleaning, data standardization, data fusion and other operations.

[0022] Step S2: using the preprocessed data as input data and using the pre-built forecasting model to predict the ship operation status of the terminal in the future, and formulating the future ship operation plan of the terminal according to the prediction results of the forecasting model.

[0023] Pre-built forecast models are models that have been pre-trained using historical terminal ship data. Input data includes the following real-time data: Meteorological data: wind speed, wind direction, temperature, humidity, air pressure, etc.

[0024] Tidal data: tide height, tidal change rate, etc.

[0025] Ship dynamic data: ship name, ship type, speed, course, estimated time of arrival (ETA), estimated time of departure (ETD), etc.

[0026] Historical operation data: the terminal’s operation records over the past period of time, including berth usage, loading and unloading efficiency, cargo types, etc.

[0027] The above pre-processed real-time data is used as input to predict the terminal ship operation situation in the future. The specific prediction content includes: Vessel Arrival Times: Forecast the estimated time of arrival (ETA) for each vessel based on AIS data and weather conditions.

[0028] Tidal Change: Predict tide heights and rates of change over the next few days based on data from tide stations.

[0029] Loading and unloading efficiency: Based on historical operation data and current weather conditions, the loading and unloading efficiency of each vessel is predicted.

[0030] Risk assessment: Combine multiple sources of data such as weather, tides, and ship dynamics to assess the risk level of future operations.

[0031] Finally, based on the output of the forecasting model, managers can develop detailed plans for future terminal vessel operations.

[0032] Step S3: Analyze the risk level of the future terminal ship operation plan to obtain a comprehensive risk score of the future terminal ship operation plan.

[0033] Step S3a, calling the input data of the current forecast model, classifying the input data according to the data type to obtain each type of input data, and arbitrarily sorting each type of input data to obtain the first to nth types of input data.

[0034] Different types of input data have different characteristics and influencing factors, so classifying them helps to better understand and process them. The classification method is as follows: Time series data: This type of data has obvious time dependence and is usually used to predict future trends. For example: Meteorological data (wind speed, wind direction, temperature, humidity, air pressure).

[0035] Tidal data (tidal height, rate of tidal change).

[0036] The time-related part of the ship's dynamic data (such as ETA, ETD).

[0037] Static data: This type of data does not change over time and is usually fixed in attributes or parameters. For example: Static information of the ship (ship name, ship type, deadweight tonnage, etc.).

[0038] Static information of berths (berth number, berth length, berth type, etc.).

[0039] Dynamic data: This type of data is updated over time and reflects the real-time status. For example: Real-time dynamic information of the ship (speed, course, current position).

[0040] Real-time operation status of the terminal (berth occupancy, loading and unloading progress).

[0041] Historical data: This type of data is a record of a period of time in the past, usually used to train models or provide reference. For example: Historical weather data.

[0042] Historical tide data.

[0043] Historical ship operation records.

[0044] In order to further optimize the structure of input data, this embodiment sorts each type of input data. The purpose of sorting is not only to make the data more orderly, but also to help the model better capture the relationship between the data. Sorting methods include but are not limited to: sorting by time, sorting by importance, sorting by data volume, and random sorting.

[0045] After the above classification and sorting, multiple categories of input data are obtained. In order to facilitate subsequent processing, these categories are numbered in sequence to form the first to nth categories of input data. Specifically: Number by data type: Assign numbers to each type of input data in turn according to the data type. For example: Category 1: Meteorological data.

[0046] Category 2: Tidal data.

[0047] Category 3: Ship dynamic data.

[0048] The fourth category: historical operation data.

[0049] Number by importance: According to the degree of influence of the data on the prediction results, the important data are ranked first and the less important data are ranked last. For example: Category 1: Vessel dynamic data (ETA, ETD, etc.).

[0050] Category 2: Meteorological data (wind speed, tide height, etc.).

[0051] Category 3: Tidal data.

[0052] The fourth category: historical operation data.

[0053] Numbering in chronological order: For time series data, you can number them according to chronological order. For example: Category 1: Meteorological data for the most recent day.

[0054] The second category: meteorological data for the past week.

[0055] The third category: meteorological data for the most recent month.

[0056] Number by data volume: According to the size of the data, the categories with smaller data volume are arranged in front, and the categories with larger data volume are arranged in the back. For example: Category 1: static ship information.

[0057] Category 2: Tidal data.

[0058] Category 3: Meteorological data.

[0059] The fourth category: historical operation data.

[0060] Step S3b: Analyze the correlation between each type of input data to obtain the correlation coefficient between each type of input data.

[0061] Different types of input data do not exist in isolation, and there may be complex interactions between them. By calculating the correlation coefficient, the relationship between these data is quantified to help better understand how different types of input data jointly affect terminal ship operations: Capturing the comprehensive impact of multiple factors: The operation of terminal ships is affected by multiple factors. Relying solely on one type of data for prediction may ignore the impact of other important factors. By analyzing the correlation between different types of data, the comprehensive impact of various factors is fully considered to improve the accuracy of the prediction.

[0062] Dealing with nonlinear relationships: The relationship between some factors may be nonlinear, such as the relationship between meteorological conditions and tidal changes. By calculating the correlation coefficient, these nonlinear relationships can be identified and the accuracy of the forecast can be further improved.

[0063] Dealing with abnormal situations: Some categories of data may have outliers or extreme situations. By analyzing the correlation coefficient, these abnormal situations can be identified in advance and corresponding measures can be taken to reduce potential risks.

[0064] Specifically: Step S3b1, obtain the actual observation value of each data in each type of input data, perform normalization on the actual observation value of each data and take the average to obtain the comprehensive observation value of each type of input data.

[0065] For example, the data on December 29, 2024 is used as the first set of observations, as follows: Weather data: Wind speed: [5 m / s, 6 m / s, 7 m / s, 8 m / s]; Wind direction: [NE, SE, SW, NW] (converted to numerical form: NE=0, SE=1, SW=2, NW=3); Temperature: [20℃, 22℃, 21℃, 23℃]; Humidity: [60%, 65%, 70%, 75%]; Air pressure: [1013 hPa, 1015 hPa, 1012 hPa, 1014 hPa]; Tidal data: Tide height: [2.5 m, 3.0 m, 3.5 m, 4.0 m]; Tidal rate of change: [0.5 m / h, 0.6 m / h, 0.7 m / h, 0.8 m / h]; Ship dynamic data: ETA: [2024-12-29 08:00, 2024-12-29 10:00, 2024-12-29 12:00, 2024-12-2914:00]; ETD: [2024-12-29 12:00, 2024-12-29 14:00, 2024-12-29 16:00, 2024-12-2918:00]; Speed: [15 knots, 16 knots, 17 knots, 18 knots]; Heading: [N, S, E, W] (converted to numerical form: N=0, S=1, E=2, W=3); Historical operation data: Berth usage: [0.8, 0.9, 0.7, 0.6]; Loading and unloading efficiency: [80%, 85%, 90%, 95%].

[0066] Each of the above actual observations is normalized using Min-Max normalization and Z-score standardization. Furthermore, meteorological data: Min-Max normalization is used. Tidal data: Z-score standardization is used. Vessel dynamic data: Min-Max normalization is used. Historical operation data: already given in percentage form, no normalization is required.

[0067] After normalization, the following results are obtained: Weather data: Normalized wind speed: [0.0, 0.33, 0.67, 1.0]; Normalized wind direction: [0.0, 0.33, 0.67, 1.0]; Normalized temperature: [0.0, 0.67, 0.33, 1.0]; Normalized humidity: [0.0, 0.33, 0.67, 1.0]; Normalized air pressure: [0.0, 0.67, 0.33, 1.0]; Tidal data: Standardized tide heights: [-1.34, -0.45, 0.45, 1.34]; Normalized tidal rate: [-1.34, -0.45, 0.45, 1.34]; Ship dynamic data: Normalized ETA: [0.0, 0.33, 0.67, 1.0]; Normalized ETD: [0.0, 0.33, 0.67, 1.0]; Normalized speed: [0.0, 0.33, 0.67, 1.0]; Normalized heading: [0.0, 0.33, 0.67, 1.0]; Historical operation data: Berth usage: [0.8, 0.9, 0.7, 0.6]; Loading and unloading efficiency: [0.8, 0.85, 0.9, 0.95].

[0068] The average of the normalized observations is taken to obtain the comprehensive observation value of this type of input data. The comprehensive observation value of the first group of observations is: meteorological data: 0.5, tidal data: 0.0, ship dynamic data: 0.5, historical operation data: 0.75.

[0069] Step S3b2, repeat step S3b1 to obtain at least two groups of comprehensive observation values ​​of each type of input data, and construct a data set of each type of input data based on the at least two groups of comprehensive observation values ​​of each type of input data.

[0070] In order to build a more reliable data set, this embodiment needs to repeat the above steps to obtain at least two sets of comprehensive observation values ​​of each type of input data. For example, the data of December 30, 2024 is selected as the second set of observation values, as follows: Weather data: Wind speed: [4 m / s, 5 m / s, 6 m / s, 7 m / s]; Wind direction: [NW, NE, SE, SW]; Temperature: [21℃, 23℃, 22℃, 24℃]; Humidity: [65%, 70%, 75%, 80%]; Air pressure: [1012 hPa, 1014 hPa, 1013 hPa, 1015 hPa]; Tidal data: Tide height: [2.0 m, 2.5 m, 3.0 m, 3.5 m]; Tidal rate of change: [0.4 m / h, 0.5 m / h, 0.6 m / h, 0.7 m / h]; Ship dynamic data: ETA: [2024-12-30 08:00, 2024-12-30 10:00, 2024-12-30 12:00, 2024-12-3014:00]; ETD: [2024-12-30 12:00, 2024-12-30 14:00, 2024-12-30 16:00, 2024-12-3018:00]; Speed: [14 knots, 15 knots, 16 knots, 17 knots]; Heading: [W, N, S, E]; Historical operation data: Berth usage: [0.7, 0.8, 0.9, 0.8]; Loading and unloading efficiency: [85%, 90%, 95%, 90%]; Repeat step S3b1 to obtain the comprehensive observation value of the second set of observation values: Meteorological data: 0.5; Tidal data: 0.0; Ship dynamic data: 0.5; Historical operation data: 0.825; In summary, this embodiment obtains two sets of comprehensive observation values ​​for each type of input data, and constructs a data set for each type of input data.

[0071] Step S3b3: Based on the data set of each type of input data, calculate the correlation coefficient between each type of input data.

[0072] The correlation coefficient between each type of input data is calculated using the Pearson correlation coefficient formula. This correlation coefficient refers to the correlation coefficient between the current type of input data and each of the remaining types of input data.

[0073] Step S3c: assigning weights to each type of input data according to the correlation coefficient of each type of input data.

[0074] Step S3c1, summing the correlation coefficients between the first category of input data and all other categories of input data to obtain the first sum of correlation coefficients; and so on, until the correlation coefficients between the nth category of input data and all other categories of input data are summed to obtain the nth sum of correlation coefficients.

[0075] The correlation coefficients between the current class input data and all the remaining class input data are summed up to obtain the first correlation coefficient sum. This is repeated until the correlation coefficients between the nth class input data and all other classes of input data are summed up to obtain the nth correlation coefficient sum.

[0076] Step S3c2, summing up all the correlation coefficients calculated in step S3c1 to obtain a total correlation coefficient.

[0077] Step S3c3: assign a weight to each type of input data according to the ratio of the sum of each correlation coefficient to the total correlation coefficient.

[0078] ; In the formula, represents the weight of the i-th type of input data, S i represents the sum of the correlation coefficients of the i-th type of input data, S t represents the overall correlation coefficient.

[0079] Step S3d: Analyze the abnormality degree of each type of input data to obtain the abnormality coefficient of each type of input data.

[0080] Abnormal data often has a greater impact on the prediction model, especially when the model does not fully consider these abnormal situations, which may lead to deviations in the prediction results. The degree of abnormality of each type of input data is analyzed.

[0081] Step S3d1: Based on the statistical analysis of historical data, a threshold interval of the actual observation value of each data in each type of input data is obtained, wherein the threshold interval includes an upper threshold limit and a lower threshold limit, and based on the upper threshold limit and the lower threshold limit, an intermediate threshold is calculated.

[0082] The threshold interval is obtained by the standard deviation or percentile method. The median of the threshold interval is used as the middle threshold.

[0083] Step S3d2: Use the threshold interval of the actual observation value of each data in each category of input data to determine the actual observation value of each data in each category of input data. If the actual observation value of the current data in each category of input data is within the threshold interval, the actual observation value of the current data is classified as a normal value; if the actual observation value of the current data in each category of input data is not within the threshold interval, the actual observation value of the current data is classified as an abnormal value.

[0084] Step S3d3: Count the number of outliers in each type of input data, and calculate the degree of deviation of each outlier from the intermediate threshold to obtain the deviation value of each outlier.

[0085] In the prior art, input data is distributed around an intermediate threshold, and most of the data points are concentrated near the intermediate threshold. Under normal operating conditions, extreme values ​​(i.e., values ​​close to or exceeding the upper and lower limits of the threshold) are usually rare. If a data point frequently appears near the upper and lower limits, this may mean that the system is experiencing some abnormal situation, or there is a problem in the data collection process. For example, in dock ship operations, parameters such as tide height and wind speed usually fluctuate within a certain range, but if the tide height at a certain moment suddenly approaches or exceeds the highest or lowest record in history, this may indicate abnormal weather or equipment failure. Rare events are often associated with abnormal events. If a data point appears outside the normal range, especially close to or exceeding the upper and lower limits of the threshold, it may be caused by some abnormal factors, so it is more likely to be an outlier. Therefore, in order to quantify the degree of abnormality of each outlier, this embodiment calculates the deviation value of each outlier from the intermediate threshold. The deviation value can be used to measure the distance of the outlier from the center of the normal range (i.e., the intermediate threshold), thereby reflecting its degree of abnormality.

[0086] Step S3d31, obtaining historical data of each type of input data, and calculating the standard deviation based on the historical data; Step S3d32, calculating the historical mean based on the historical data; Step S3d33, calculating the deviation value of each outlier according to the standard deviation, the historical mean, the actual observed value of the outlier, the intermediate threshold, the upper threshold, and the lower threshold; The calculation formula of the deviation value is: ; Where D i Represents the deviation value of the i-th type of input data, x (i,j) represents the actual observed value of the jth outlier in the i-th input data, T m represents the middle threshold, U represents the upper threshold, and L represents the lower threshold. represents the weight coefficient, n represents the total number of categories of input data, and m represents the total number of outliers. represents the historical mean, Represents standard deviation.

[0087] , which is the absolute difference, measures the relative distance between the outlier and the middle threshold, and is standardized as a multiple of the standard deviation. This allows data of different dimensions to be compared, and can reflect the degree of deviation of the outlier from the center of the normal data distribution. A larger absolute difference indicates that the outlier is farther from the center of the normal range and is more abnormal.

[0088] , when the outlier is close to the mean, this part should be close to 1 / 2. When the outlier is much larger than the mean, the fractional part of e is a large positive number, and e and its fractional power are very small positive numbers close to 0, making this part tend to 1. When the outlier is much smaller than the mean, the fractional part of e is a large negative number, the fractional power of e is very large, and the whole part is close to 0. It helps to smoothly handle extreme values, avoid situations where the deviation is too large or too small, and better capture the gradual changes of outliers.

[0089] , which measures the minimum distance of the outlier from the upper and lower thresholds and normalizes it as a ratio of the threshold interval width. Outliers close to the upper and lower limits will get higher deviation values ​​because they are more likely to be abnormal. This ensures that data points that are close to or outside the upper and lower thresholds can be effectively identified, not just those points far from the middle threshold.

[0090] In summary, not only the distance between the outliers and the center of normal data distribution is considered, but also the degree of proximity between the outliers and the upper and lower limits of the threshold is paid special attention to. Such a design can more comprehensively evaluate the degree of deviation of outliers, capturing the impact of extreme values ​​while considering the distribution characteristics of data. Overall, this design can provide more accurate and reliable anomaly detection results in a variety of application scenarios, helping to identify potential risks and problems.

[0091] Step S3d4: sum the deviation values ​​of each abnormal value in each type of input data to obtain the abnormal coefficient of each type of input data.

[0092] Step S3e: comprehensively consider the correlation coefficient between each type of input data, the weight of each type of input data, and the abnormal coefficient of each type of input data to calculate the risk score of each type of input data.

[0093] ; In the formula, R i represents the risk score of the i-th type of input data, represents the weight of the i-th type of input data, S i represents the sum of the correlation coefficients of the i-th type of input data, Y i represents the abnormal coefficient of the i-th type of input data, Represents the adjustment factor.

[0094] Step S3f, summing up the risk scores of each type of input data to obtain a comprehensive risk score of the future terminal ship operation plan.

[0095] ; In the formula, R s represents the comprehensive risk score, R irepresents the risk score of the i-th category of input data, and n represents the total number of categories of input data.

[0096] By summing the risk scores of each type of input data, the comprehensive risk score of the future terminal ship operation plan can be obtained, which can achieve a comprehensive risk assessment of the entire operation system. This method can not only quantify the abnormalities of various types of input data, but also integrate these abnormalities into a unified risk indicator, helping managers to more intuitively understand the safety and potential risks of the current operating environment. Through the comprehensive risk score, terminal managers can promptly identify which factors may have an adverse impact on ship operations and take corresponding preventive measures, such as adjusting the ship's entry and exit time, increasing the frequency of safety inspections, or launching emergency plans. In addition, the comprehensive risk score can also provide data support for long-term planning, help optimize terminal resource allocation and operation processes, and improve overall operational efficiency. Overall, this comprehensive scoring mechanism can improve the flexibility and response speed of terminal ship operations, reduce the accident rate, and ensure the stable operation of the port while ensuring safety.

[0097] Step S4: set a warning threshold for the comprehensive risk score. If the current comprehensive risk score is greater than or equal to the warning threshold, an alarm is issued, and preventive measures for future terminal ship operation plans are formulated based on the current comprehensive risk score; if the current comprehensive risk score is less than the warning threshold, no alarm is required.

[0098] Collect historical terminal ship data in the past period of time and the corresponding future terminal ship operation conditions, calculate the comprehensive risk score for the historical terminal ship data, and obtain the warning threshold based on the mean and standard deviation of the calculated comprehensive risk score. For example, the low risk warning threshold is 50 points, the medium risk warning threshold is 70 points, and the high risk warning threshold is 85 points; The risk level of the future terminal ship operation plan predicted in the above steps is determined according to the early warning threshold. Preventive measures are formulated according to the risk level. For example: 1. The comprehensive risk score is 75 points, which is medium risk. The preventive measures are: Adjust operation plans: According to current weather conditions and tidal changes, appropriately adjust the time of ships entering and leaving the port to avoid operations during periods of high wind speed or low tide.

[0099] Increased safety inspections: Carry out additional safety inspections of terminal facilities and vessels to ensure that all equipment is in good condition.

[0100] Strengthen monitoring: Increase the number of on-site monitoring personnel, track the progress of operations in real time, and handle emergencies in a timely manner.

[0101] Prepare emergency plans: Prepare emergency rescue equipment and personnel to ensure rapid response in emergency situations.

[0102] 2. The comprehensive risk score is 60 points, which is low risk. The preventive measures are: Routine monitoring: Maintain daily monitoring and inspections to ensure a safe working environment.

[0103] Regular maintenance: Perform regular maintenance of equipment as planned to ensure the normal operation of the facility.

[0104] 3. The comprehensive risk score is 90 points, which is high risk. The preventive measures are: Suspend operations: Temporarily stop the entry and exit of ships into and out of the port, and resume them after the risk is reduced.

[0105] Evacuate personnel: Based on the actual situation, organize the dock staff to evacuate to a safe area.

[0106] Activate the emergency plan: fully activate the emergency plan, mobilize emergency rescue forces, and ensure the safety of people and property.

[0107] Strengthen communication: Maintain close contact with the meteorological department, maritime administration and other relevant agencies to obtain the latest weather forecasts and navigation information, and adjust response strategies in a timely manner.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A method for forecasting and warning of ship operations at a dock, characterized in that: The following steps are involved: Step S1, collecting terminal ship data through at least two sources, and performing preprocessing on the data; Step S2: using the preprocessed data as input data and using the pre-built forecasting model to forecast the ship operation situation at the terminal in the future, and formulating a future ship operation plan at the terminal according to the forecasting results of the forecasting model; Step S3, analyzing the risk level of the future terminal ship operation plan to obtain a comprehensive risk score of the future terminal ship operation plan; Step S3a, calling the input data of the current forecast model, classifying the input data according to the data type to obtain each type of input data, and arbitrarily sorting each type of input data to obtain the first to nth types of input data; Step S3b, analyzing the correlation between each type of input data to obtain the correlation coefficient between each type of input data; Step S3c, assigning a weight to each type of input data according to the correlation coefficient of each type of input data; Step S3d, analyzing the abnormality degree of each type of input data to obtain the abnormality coefficient of each type of input data; Step S3d1, based on the statistical analysis of historical data, a threshold interval of the actual observation value of each data in each type of input data is obtained, wherein the threshold interval includes an upper threshold limit and a lower threshold limit, and based on the upper threshold limit and the lower threshold limit, an intermediate threshold value is calculated; Step S3d2, using the threshold interval of the actual observation value of each data in each type of input data to determine the actual observation value of each data in each type of input data, if the actual observation value of the current data in each type of input data is within the threshold interval, the actual observation value of the current data is classified as a normal value; If the actual observation value of the current data in each type of input data is not within the threshold interval, the actual observation value of the current data is classified as an outlier; Step S3d3, counting the number of outliers in each type of input data, and calculating the degree of deviation of each outlier from the intermediate threshold, to obtain the deviation value of each outlier; Step S3d4, summing the deviation values ​​of each abnormal value in each type of input data to obtain the abnormal coefficient of each type of input data; Step S3e, comprehensively analyzing the correlation coefficient between each type of input data, the weight of each type of input data, and the abnormal coefficient of each type of input data, and calculating the risk score of each type of input data; Step S3f, summing the risk scores of each type of input data to obtain a comprehensive risk score of the future terminal ship operation plan; Step S4: setting a warning threshold for the comprehensive risk score. If the current comprehensive risk score is greater than or equal to the warning threshold, an alarm is issued, and preventive measures for future terminal ship operation plans are formulated based on the current comprehensive risk score. If the current comprehensive risk score is less than the warning threshold, there is no need to issue an alarm.

2. A method for forecasting and warning of dock ship operations according to claim 1, characterized in that: Calculate the degree of deviation of each outlier from the middle threshold to obtain the deviation value of each outlier, including: Step S3d31, obtaining historical data of each type of input data, and calculating the standard deviation based on the historical data; Step S3d32, calculating the historical mean based on the historical data; Step S3d33, calculate the deviation value of each outlier according to the standard deviation, mean, actual observed value of the outlier, intermediate threshold, upper threshold, and lower threshold.

3. A method for forecasting and warning of ship operations at a dock according to claim 2, characterized in that: The calculation formula of the deviation value is: ； Where D i represents the deviation value of the i-th type of input data, x (i,j) represents the actual observed value of the jth outlier in the i-th input data, T m represents the middle threshold, U represents the upper threshold, and L represents the lower threshold. represents the weight coefficient, n represents the total number of categories of input data, and m represents the total number of outliers. represents the historical mean, Represents standard deviation.

4. A method for forecasting and warning of ship operations at a dock according to claim 1, characterized in that: The correlation between each type of input data is analyzed to obtain the correlation coefficient between each type of input data, including: Step S3b1, obtaining the actual observation value of each data in each type of input data, normalizing the actual observation value of each data and taking the average, to obtain the comprehensive observation value of each type of input data; Step S3b2, repeating step S3b1 to obtain at least two groups of comprehensive observation values ​​of each type of input data, and constructing a data set of each type of input data based on the at least two groups of comprehensive observation values ​​of each type of input data; Step S3b3: Based on the data set of each type of input data, calculate the correlation coefficient between each type of input data.

5. A method for forecasting and warning of ship operations at a dock according to claim 1, characterized in that: Each type of input data is weighted according to its correlation coefficient, including: Step S3c1, summing the correlation coefficients between the first category of input data and the input data of all other categories to obtain a first sum of correlation coefficients; and so on, until the correlation coefficients between the nth category of input data and the input data of all other categories are summed to obtain an nth sum of correlation coefficients; Step S3c2, summing up all the correlation coefficients calculated in step S3c1 to obtain a total correlation coefficient; Step S3c3: assign a weight to each type of input data according to the ratio of the sum of each correlation coefficient to the total correlation coefficient.

6. A method for forecasting and warning of ship operations at a dock according to claim 5, characterized in that: The calculation formula for the weight of each type of input data is: ； In the formula, represents the weight of the i-th type of input data, S i represents the sum of the correlation coefficients of the i-th type of input data, S t represents the overall correlation coefficient.

7. A method for forecasting and warning of ship operations at a dock according to claim 1, characterized in that: The risk score for each type of input data is calculated as: ； In the formula, R i represents the risk score of the i-th type of input data, represents the weight of the i-th type of input data, S i represents the sum of the correlation coefficients of the i-th type of input data, Y i represents the abnormal coefficient of the i-th type of input data, Represents the adjustment factor.

8. A method for forecasting and warning of ship operations at a dock according to claim 1, characterized in that: The formula for calculating the comprehensive risk score is: ； In the formula, R s represents the comprehensive risk score, R i represents the risk score of the i-th category of input data, and n represents the total number of categories of input data.

Images