Seaman workload adaptive prediction method based on multi-source heterogeneous data fusion

Abstract

The invention discloses a sailor workload adaptive prediction method based on multi-source heterogeneous data fusion. The method comprises the following steps: acquiring physiological feature data of sailors to obtain the current physiological load of an individual; obtaining an individual physiological load prediction value in the next time period based on the current physiological load of the individual and the physiological load parameters at the historical moment; acquiring the navigation environment data of the ship where the crew is located in the next time period, and obtaining the comprehensive pressure level of the navigation environment where the crew is located in the next time period; and inputting a causal reasoning graph model based on the comprehensive pressure level of the sailing environment in which the sailors are located in the next time period and the individual physiological load prediction value in the next time period, and predicting the workload level of the sailors. According to the method, the physiological feature data, the historical physiological load parameters and the navigation environment data of the sailors are collected and fused, the working states of the sailors are comprehensively reflected in multiple dimensions, and the information amount and the comprehensive analysis capacity of the prediction model are improved.

Description

Technical Field

[0001] The present invention belongs to the field of computer technology, and in particular relates to a crew workload adaptive prediction method based on multi-source heterogeneous data fusion. Background Art

[0002] Crew members undertake complex and changeable tasks during ship operation, and the reasonable management of their workload is of great significance to ensure the safe and efficient operation of the ship. Traditional crew workload assessment methods mainly rely on a single data source or a static assessment model, but these methods have many shortcomings.

[0003] First, it is impossible to fully reflect the actual workload status of crew members in different working environments. During the voyage of a ship, the workload of crew members is not only affected by their own physiological state, but also by the combined effects of multiple factors such as the dynamically changing navigation environment and multi-task parallelism. A single data source is difficult to capture these complex influencing factors.

[0004] In addition, there are significant differences among crew members in their physiological characteristics, work habits and ability to cope with stress. Traditional methods often use unified assessment criteria and are unable to conduct accurate assessments of different individuals.

[0005] In response to the above problems, there is an urgent need for a crew workload prediction method that can integrate multi-source heterogeneous data and achieve individualization and dynamic adaptation, so as to improve the comprehensiveness, accuracy and practicality of the prediction, thereby effectively supporting ship operation management and ensuring the health and safety of crew members. Summary of the invention

[0006] In view of the defects existing in the above-mentioned prior art, the present invention provides a crew workload adaptive prediction method based on multi-source heterogeneous data fusion, comprising:

[0007] Collecting physiological characteristic data of the crew members, inputting it into the individual physiological load model P, and obtaining the current physiological load of the individual;

[0008] Based on the current physiological load of the individual and the physiological load parameters at historical moments, inputting into a time series prediction model to obtain a predicted value of the individual physiological load in the next time period;

[0009] Obtain the navigation environment data of the crew's ship in the next time period;

[0010] Based on the navigation environment data of the next time period, input the environmental assessment model E to obtain the comprehensive stress level of the navigation environment in which the crew is located in the next time period;

[0011] Based on the comprehensive stress level of the sailing environment in which the crew members are located in the next time period and the predicted value of the individual physiological load in the next time period, the causal reasoning graph model C is input to predict the workload level of the crew members.

[0012] Among them, wearable devices are used to collect physiological characteristic data of crew members in real time, including:

[0013] Heart rate HR(t), heart rate variability HRV(t), activity measured by accelerometer Activity(t), skin temperature Temp(t), skin conductivity EDA(t).

[0014] Among them, the individual physiological load model P is defined, and the individual physiological load model P is used to calculate the physiological load L(t) of the crew at the current time t; the model considers the nonlinear accumulation of fatigue and the physiological dynamic recovery process, and the formula is as follows:

[0015]

[0016] in,

[0017] Lt(t): physiological load level at the current time t;

[0018] L(t-Δt): physiological load level at the previous moment t-Δt;

[0019] HR(t): heart rate measured at the current time t;

[0020] HR rest : individual's resting heart rate;

[0021] HRV(t): heart rate variability measured at the current time t;

[0022] HRV rest : Individual resting heart rate variability;

[0023] Activity(t): the amount of activity at the current time t;

[0024] Δt: time interval;

[0025] α1, α2, α3: individualized weight coefficients;

[0026] n1, n2, n3: nonlinear index;

[0027] γ: Physiological load recovery rate parameter.

[0028] Among them, based on the individual's current physiological load and the physiological load parameters at historical moments, the individual's physiological load in the next time period is predicted, including:

[0029] When predicting the physiological load value of the next time period, the current and historical physiological load data, as well as the model parameters, are used to predict the physiological load of the time period (t+Δt) The navigation environment data of the crew's ship in the next time period is obtained through the AIS system or the meteorological and sea condition service channel, and based on this, the environmental assessment model E is input to obtain the comprehensive stress level of the crew's navigation environment in the next time period.

[0030] The environmental assessment model E is defined, and the environmental assessment model E is used to convert the above-obtained navigation environment data into the comprehensive stress level S of the navigation environment in which the crew is located. env :

[0031] S env =E(environmental parameter)=β1S weather +β2S sea +β3S nav +β4S traffic ;

[0032] in,

[0033] S env : Comprehensive environmental stress level;

[0034] Weight coefficient β i : represents the weight of each environmental factor on the comprehensive pressure, satisfying β1+β2+β3+β4=1, and β i ≥0;

[0035] Among them, the environmental pressure sub-indicators include:

[0036] Weather Pressure Index S weather :S weather =w1f wind (V w )+w2f visibibity (Vis)+w3f rain (Rain)+w4f temp (Temp);

[0037] Sea pressure index S sea :S sea =w5f wave (H wave )+w6f current (V c );

[0038] Navigation stress index

[0039] S nav :S nav =w7f route (R complexity )+w8f maneuver(M frequency );

[0040] Traffic Pressure Index

[0041] S traffic :S traffic =w9f vesseldensity (D vessel )+w 10 f collisionrisk (C risk );

[0042] in,

[0043] Weight coefficient w i : represents the influence weight of each specific factor on the corresponding pressure index, satisfying that the sum of the weight coefficients in the corresponding pressure index is 1, and w i ≥0;

[0044] Function f converts environmental variables into standardized stress scores, ranging from 0 to 1.

[0045] Among them, based on the comprehensive stress level of the crew's navigation environment in the next time period and the predicted value of individual physiological load, the causal reasoning graph model C is input to predict the crew's workload level in the future, including:

[0046] Construct a causal reasoning graph model C to describe the causal relationship between variables;

[0047] Based on regression model, we capture the effects of physiological load and environmental stress on workload, as well as their interaction;

[0048] Use parameter estimation methods to estimate model parameters.

[0049] Among them, constructing the causal reasoning graph model C includes the following processes:

[0050] First, the variable definitions:

[0051] : The predicted value of individual physiological load in the next time period t+Δt;

[0052] S env (t+Δt): the comprehensive pressure level of the navigation environment in the next time period t+Δt;

[0053] W(t+Δt): crew workload level in the future;

[0054] Determine causal relationships, including defining individual physiological loads And the combined pressure of the navigation environment S env (t+Δt) has a direct impact on the crew workload level W(t+Δt).

[0055] Wherein, the regression model can be expressed as:

[0056]

[0057] in:

[0058] W(t+Δt): predicted value of workload level;

[0059] β0: bias term, indicating the basic workload level;

[0060] β1, β2, β3: model coefficients, indicating the influence of the corresponding variables on the workload level;

[0061] :Individual physiological load prediction value;

[0062] S env (t+Δt): comprehensive pressure level of navigation environment;

[0063] n1, n2, n3: nonlinear index, controlling the nonlinearity of the variable;

[0064] In order to estimate the model parameters θ = {β0, β1, β2, β3, n1, n2, n3}, a nonlinear least squares method is used to minimize the error through an iterative optimization algorithm.

[0065] Wherein, based on the estimated rest time of the next time period obtained from the work and rest schedule, the initially predicted workload level is corrected for the first time;

[0066] Based on the environmental factors of the route and the personnel situation in the work and rest arrangement of the next period, the workload level after the first revision is revised for the second time;

[0067] The intermediate values ​​between two correction processes and the workload level before each correction are recorded to obtain the multi-stage, multi-factor load and the relationship between the corrected load and the cause and feed it back to the control center.

[0068] The present invention utilizes the individual physiological load model P and builds a model according to the physiological data of each crew member, thereby achieving accurate assessment of the workload of different individuals and enhancing the individual adaptability and prediction accuracy of the model.

[0069] Based on current and historical physiological load data, the present invention dynamically predicts the physiological load of an individual in a future period of time through a time series prediction model to ensure the real-time and forward-looking nature of the prediction results.

[0070] The present invention introduces an environmental assessment model E to systematically assess the comprehensive pressure level of the navigation environment in the next time period, quantitatively incorporates environmental factors into workload prediction, and improves the sensitivity and responsiveness of the model to changes in the external environment.

[0071] The present invention organically combines the predicted value of individual physiological load with the environmental stress level through the causal reasoning graph model C, deeply explores the causal relationship between various factors, and improves the interpretability and accuracy of workload prediction.

[0072] The present invention comprehensively utilizes the synergy of multi-source data and multiple models to achieve adaptive prediction of crew workload. It can dynamically adjust the prediction results according to real-time data and environmental changes to ensure that the crew workload is always within a controllable range under various complex navigation conditions.

[0073] The present invention collects and integrates the crew's physiological characteristic data, historical physiological load parameters and navigation environment data to comprehensively and multi-dimensionally reflect the crew's working status, thereby improving the information content and comprehensive analysis capabilities of the prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present disclosure will become readily understood. In the accompanying drawings, several embodiments of the present disclosure are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0075] Figure 1 The invention is a flowchart showing a crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to an embodiment of the present invention. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two.

[0078] It should be understood that although the terms first, second, third, etc. may be used to describe ... in the embodiments of the present invention, these ... should not be limited to these terms. These terms are only used to distinguish .... For example, without departing from the scope of the embodiments of the present invention, the first ... may also be referred to as the second ..., and similarly, the second ... may also be referred to as the first ....

[0079] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0080] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0081] It should also be noted that the term "includes", "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, so that a commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the commodity or device including the elements.

[0082] Existing methods fail to fully consider the physiological differences of individual crew members and their dynamic changes, resulting in a lack of individualized and real-time adaptability in the prediction results, and an inability to timely reflect the real-time changes in the crew status. During the voyage of the ship, environmental factors (such as weather, sea conditions, etc.) have a significant impact on the crew's workload. Existing methods fail to systematically and comprehensively evaluate the comprehensive impact of environmental pressure on workload in the prediction. Existing prediction models often lack in-depth modeling of the causal relationship between various influencing factors, resulting in insufficient interpretability and accuracy of the prediction results.

[0083] like Figure 1 As shown, the present invention discloses a crew workload adaptive prediction method based on multi-source heterogeneous data fusion, comprising the following steps:

[0084] Using wearable devices such as smart watches to collect physiological characteristic data of crew members, inputting it into the individual physiological load model P, and obtaining the current physiological load of the individual;

[0085] Based on the current physiological load of the individual and the physiological load parameters at historical moments, inputting into a time series prediction model to obtain a predicted value of the individual physiological load in the next time period;

[0086] Obtain the navigation environment data of the individual ship in the next time period through the AIS system or meteorological and sea condition service channels;

[0087] Based on the navigation environment data of the next time period, input the environmental assessment model E to obtain the comprehensive stress level of the navigation environment in which the crew is located in the next time period;

[0088] Based on the comprehensive stress level of the sailing environment in which the crew members are located in the next time period and the predicted value of the individual physiological load in the next time period, the causal reasoning graph model C is input to predict the workload level of the crew members in the future period.

[0089] The present invention utilizes the individual physiological load model P and builds a model according to the physiological data of each crew member, thereby achieving accurate assessment of the workload of different individuals and enhancing the individual adaptability and prediction accuracy of the model.

[0090] Based on current and historical physiological load data, the present invention dynamically predicts the physiological load of an individual in the next period of time through a time series prediction model to ensure the real-time and forward-looking nature of the prediction results. The present invention introduces an environmental assessment model E to systematically assess the comprehensive pressure level of the navigation environment in the next period of time, quantitatively incorporates environmental factors into workload prediction, and improves the model's sensitivity and responsiveness to changes in the external environment.

[0091] The present invention organically combines the predicted value of individual physiological load with the environmental stress level through the causal reasoning graph model C, deeply explores the causal relationship between various factors, and improves the interpretability and accuracy of workload prediction.

[0092] The present invention comprehensively utilizes the synergy of multi-source data and multiple models to achieve adaptive prediction of crew workload. It can dynamically adjust the prediction results according to real-time data and environmental changes to ensure that the crew workload is always within a controllable range under various complex navigation conditions.

[0093] In one embodiment, wearable devices such as smart watches are used to collect physiological characteristic data of crew members, which are input into individual physiological load model P to obtain the individual's current physiological load, including the physiological dynamic accumulation process of fatigue based on the fatigue accumulation formula. The following is a specific implementation method: In one embodiment, wearable devices such as smart watches are used to collect physiological characteristic data of crew members in real time, including:

[0094] Heart Rate (HR(t))

[0095] Heart Rate Variability (HRV(t))

[0096] Activity (measured by accelerometer, Activity(t))

[0097] Skin temperature (Temp(t))

[0098] Skin conductivity (EDA(t))

[0099] Define the individual physiological load model P, which is used to calculate the physiological load L(t) of the crew at the current time t. The model considers the nonlinear accumulation of fatigue and the physiological dynamic recovery process, and the formula is as follows:

[0100]

[0101] in,

[0102] L(t): Physiological load level (fatigue level) at the current time t.

[0103] L(t-Δt): Physiological load level at the previous moment t-Δt.

[0104] HR(t): Heart rate measured at the current time t.

[0105] HR rest : The individual's resting heart rate (baseline heart rate).

[0106] HRV(t): Heart rate variability measured at the current time t.

[0107] HRV rest : Individual's resting heart rate variability (baseline HRV).

[0108] Activity(t): The amount of activity at the current time t (obtained through the accelerometer).

[0109] Δt: time interval (the time difference between two sampling moments).

[0110] α1, α2, α3: Individualized weight coefficients, reflecting the influence of each physiological index on fatigue accumulation. n1, n2, n3: Nonlinear index, used to describe the nonlinear relationship between physiological index and fatigue accumulation. γ: Physiological load recovery rate parameter, reflecting the speed of fatigue recovery over time.

[0111] Among them, the sub-item Heart rate contribution: Indicates the contribution of the current heart rate higher than the resting heart rate to fatigue. The index n1 reflects nonlinearity. Indicates the contribution of reduced heart rate variability to fatigue. The lower the HRV, the higher the fatigue. Activity Contribution Indicates the contribution of physical activity to fatigue. Fatigue recovery part L(t-Δt)×e -γΔt Use the exponential decay function e -γΔt Simulate the recovery process of fatigue over time. The larger γ is, the faster the fatigue recovery is.

[0112] Among them, the baseline parameter HRR rest and HRV rest It needs to be obtained through multiple measurements in a resting state to reflect the individual's physiological baseline level. The weight coefficients and indexes α1, α2, α3, n1, n2, and n3 can be individually fitted through regression analysis of historical data or machine learning methods to accurately reflect the sensitivity of each crew member to different physiological indicators. The recovery rate parameter γ can be adjusted according to the individual's fatigue recovery characteristics.

[0113] Exemplarily, the model parameters are as follows: α1=0.5, n1=2; α2=0.3, n2=1;

[0114] α3=0.2, n3=1.5; γ=0.1.

[0115] By setting the indexes n1, n2, and n3, the model can reflect the nonlinear effect of physiological indicators on fatigue accumulation. For example, when the heart rate exceeds a certain threshold, the fatigue accumulation rate accelerates. The model parameters are adjusted based on individual historical data to reflect the physiological characteristics and fatigue sensitivity of each crew member. The model also takes into account the accumulation and recovery process of fatigue, and can reflect the physiological load status of the crew in real time.

[0116] In one embodiment, an initial fatigue level L(0)=L0 is set, which can usually be set to zero or a baseline value. At each time interval Δt, a new physiological load L(t) model is calculated based on the latest physiological characteristic data to continuously accumulate fatigue in time while considering the recovery process of fatigue.

[0117] In one embodiment, based on the individual's current physiological load and the physiological load parameters at historical moments, the individual's physiological load in the next time period is predicted, including using the physiological load L(t-iΔt) at the past p moments to predict the current physiological load L(t), and the error term ∈(t-jΔt) at the past q moments to correct the prediction:

[0118]

[0119] Specifically, when predicting the physiological load value of the next time period, the current and historical physiological load data, as well as the model parameters, are used to predict the physiological load of the next time period (t+Δt) The specific prediction method is:

[0120]

[0121] Since t+Δt-iΔt=t-(i-1)Δt,

[0122]

[0123] The parameters are defined as follows:

[0124] L(t): The physiological load of an individual at time t.

[0125] Δt: time interval, the time difference between two sampling moments.

[0126] p: indicates how many past physiological load data are used.

[0127] q: indicates how many past error terms to use.

[0128] : Autoregressive coefficient, i = 1, 2, ..., p, indicating the influence of past physiological load on current value. θ j : Sliding mean coefficient, j = 1, 2, …, q, which indicates the influence of past errors on the current value.

[0129] ∈(t): Random error term (noise), assumed to be white noise with zero mean.

[0130] The future error term ∈(t+Δt) is unknown and its expectation is usually assumed to be zero, so the future error term is not included in the forecast.

[0131] In one embodiment, the model parameter estimation includes the following process: collecting physiological load time series data L(t) of sufficient length and calculating the corresponding error term ∈(t), the error term calculation is: Use historical data to estimate model coefficients using methods such as least squares or maximum likelihood estimation and θ j .

[0132] In one embodiment, the historical data collected includes physiological load data of the past p moments: L(t), L(t-Δt), L(t-2Δt),…, L(t-(p-1)Δt), and error terms of the past q moments: ∈(t), ∈(t-Δt), ∈(t-2Δt),…, ∈(t-(q-1)Δt).

[0133] Calculate predicted values and Predict the physiological load at the next moment:

[0134] In one embodiment, it is assumed that: model order: p=2, q=1.

[0135] Estimated model coefficients: θ1=0.5.

[0136] Time interval: Δt = 1 (unit time, for example 1 hour)

[0137] Time interval: Δt = 1 (unit time, for example 1 hour)

[0138] Physiological load: L(t)=25, L(t-1)=23.

[0139] Error terms: ∈(t)=0.8, ∈(t-1)=-0.5.

[0140] Prediction calculation: θ j ∈(t-(j-1)Δt)=0.5×0.8=0.4.

[0141] Predict the physiological load at the next moment: Therefore, the physiological load in the next time period (t+Δt) is predicted to be 22.5.

[0142] By inputting the individual's current physiological load L(t) and its historical physiological load data and error terms into the prediction model, the individual's physiological load in the next time period can be predicted. In one embodiment, the navigation environment data of the individual ship in the next time period is obtained through the AIS system or the meteorological and sea condition service channel, and based on this input into the environmental assessment model E, the comprehensive stress level of the navigation environment of the crew in the next time period is obtained.

[0143] In one embodiment, obtaining navigation environment data includes:

[0144] Obtain navigation related data through the AIS system:

[0145] AIS (Automatic Identification System) provides real-time dynamic information of ships at sea, including ship information: ship position (latitude and longitude) (Lat, Lon)

[0146] Ship speed (speed over ground) V s

[0147] Ship heading (relative to true north) H s

[0148] Vessel status (underway, anchored, etc.)

[0149] Other ship information: Number of nearby ships N v

[0150] Distance D of nearby ships i (i=1,2,…,N v ) The course and speed of other ships and environmental data obtained through meteorological and sea condition service channels, including meteorological data and sea condition data:

[0151] Wind speed V w

[0152] Wind direction H w

[0153] Visibility

[0154] Rainfall

[0155] Temperature

[0156] Wave height H wave

[0157] Wave direction (wave propagation direction) D wave

[0158] Flow rate V c

[0159] Flow to D c

[0160] Among them, all the data obtained should be for the next time period, that is, the forecast data within the time Δt starting from the current time t.

[0161] In one embodiment, an environmental assessment model E is defined, and the environmental assessment model E is used to convert the above-obtained navigation environment data into a comprehensive stress level S of the navigation environment in which the crew is located. env .

[0162] The model reflects the impact of various environmental factors on crew stress:

[0163] S env =E(environmental parameter)=β1S weather +β2S sea +β3s nav +β4S traffic .

[0164] in,

[0165] S env : Comprehensive environmental pressure level (numerical pressure index).

[0166] Weight coefficient β i : represents the weight of each environmental factor on the comprehensive pressure, satisfying β1+β2+β3+β4=1, and β i ≥0.

[0167] Among them, the environmental pressure sub-indicators include:

[0168] Weather Pressure Index S weather :S weather =w1f wind (V w )+w2f visibibity (Vis)+w3f rain (Rain)+w4f temp (Temp).

[0169] Sea pressure index S sea :S sea =w5f wave (H wave )+w6f current (V c ).

[0170] Navigation stress index

[0171] S nav :S nav =w7f route (R complexity )+w8f maneuver (M frequency ).

[0172] Traffic Pressure Index

[0173] S traffic :S traffic =w9f vesseldensity (D vessel )+w 10 f collisionrisk (C risk ).

[0174] in,

[0175] Weight coefficient w i : represents the influence weight of each specific factor on the corresponding pressure index, satisfying that the sum of the weight coefficients in the corresponding pressure index is 1, and w i ≥0.

[0176] Function f: Converts specific environmental variables into standardized stress scores (0 to 1).

[0177] Among them, the weather pressure index S weather ,include:

[0178] Wind speed influence function f wind (V w ): As wind speed increases, pressure increases and can be defined as:

[0179]

[0180] in:

[0181] V w : Wind speed (unit: m / s).

[0182] : Maximum wind speed, corresponding to the wind speed threshold of pressure score 1.

[0183] Visibility influence function f visibility (Vis): The lower the visibility, the greater the pressure, which can be defined as:

[0184] in:

[0185] Vis: visibility (unit: km).

[0186] Vis max : Ideal visibility with minimal stress.

[0187] Viss min : Lowest visibility, highest pressure.

[0188] Rainfall impact function f rain (Rain): The greater the rainfall, the greater the pressure. The function can be defined as:

[0189]

[0190] in:

[0191] Rain: rainfall (unit: mm / h).

[0192] Rain max : Maximum rainfall value, corresponding to the rainfall threshold with a pressure score of 1.

[0193] Temperature influence function f temp (Temp): Too high or too low temperature will increase pressure, which can be expressed as:

[0194] in:

[0195] Temp: Temperature.

[0196] Temp opt : The most suitable temperature and the lowest pressure.

[0197] Temp range : Half-width of the temperature range, used for normalization.

[0198] Among them, the sea pressure index S sea ,include:

[0199] Wave height influence function fwave (H wave ): The higher the wave height, the greater the pressure, which can be defined as:

[0200] in:

[0201] H wave : Wave height (unit: m).

[0202] : Maximum wave height, corresponding to the threshold of pressure score 1.

[0203] Flow velocity influence function f current (V c ): The greater the flow rate, the greater the pressure, expressed as:

[0204]

[0205] in:

[0206] V c : Flow rate (unit: m / s).

[0207] : Maximum flow rate, corresponding to the threshold of pressure score 1.

[0208] Among them, the navigation pressure index S nav include:

[0209] Route complexity influence function f route (R complexity ): The more complex the route, the greater the pressure, expressed as:

[0210]

[0211] in:

[0212] R complexity Score the complexity of the route (0 to R max ).

[0213] R max : The maximum value of the route complexity score.

[0214] The manipulation frequency affects the function f maneuver (M frequency ): When frequent manipulation is required, the pressure increases, expressed as:

[0215] in:

[0216] M frequency : The expected number of operations in the next time period.

[0217] M max: Maximum expected number of operations.

[0218] Among them, the operating frequency M frequency The types of maneuvers include all operations that require the crew to actively control the ship to change its course, speed or position, such as turning, slowing down or accelerating, changing the course, avoiding other ships, and emergency maneuvers to deal with unexpected situations.

[0219] M frequency The determination is usually based on the following factors:

[0220] Route complexity: the number of turns, crossings, narrow waters, shallow waters, etc. required for maneuvering during the route.

[0221] Traffic flow: When the density of surrounding vessels is high, the need for maneuvering to avoid other vessels increases.

[0222] Weather and sea conditions: In adverse weather conditions, more maneuvers may be required to maintain safe navigation.

[0223] Voyage plan: The estimated number of maneuvers required based on the intended route and schedule.

[0224] Determine M frequency The process includes the following:

[0225] Step 1: Assess route complexity

[0226] Get route information: including distance, route type, number of turning points, etc.

[0227] Calculate the number of turns N of the route turns : Count the number of turning points in the route.

[0228] Step 2: Assess traffic conditions

[0229] Get AIS data: Get the number and distribution of other ships in the navigation area in the next time period.

[0230] Calculate the number of avoidance requirements N avoid : Estimate the number of times you will need to make avoidance based on the number of other vessels and the expected encounters.

[0231] Step 3: Assess the impact of weather and sea conditions

[0232] Get weather and sea condition forecasts: including wind, waves, currents and more.

[0233] Evaluate the increase in control demand ΔN weather :The number of operations that may need to be increased under adverse conditions. Step 4: Calculate the total number of operations M frequency =N turns +N avoid +ΔN weather , where N turns: Number of turns in the route; N avoid : The estimated number of times you need to avoid other ships; ΔN weather : Increased number of maneuvers due to weather and sea conditions.

[0234] For example, assume: Route information: The route length is 100 nautical miles and contains 5 turning points.

[0235] Traffic situation: It is expected that within the next time period, three other ships will be encountered that need to be avoided.

[0236] Weather and sea conditions: Weather forecast shows strong winds and waves, requiring 2 additional maneuvers to maintain course and speed.

[0237] Calculation: N turns =5; N avoid =3; ΔN weather =2. Total number of operations

[0238] Number:M frequency =5+3+2=10.

[0239] Traffic Pressure Index S traffic ,include:

[0240] Ship density influence function f vesseldensity (D vessel ): The greater the density of nearby ships, the greater the pressure.

[0241] It is expressed as:

[0242] in:

[0243] D vessel : Number of ships per unit area.

[0244] D max : Maximum value of ship density.

[0245] Collision risk impact function f collisionrisk (C risk ): The higher the risk of collision with other ships, the greater the pressure, expressed as: f collisionrisk (C risk )=min(C risk ,1),

[0246] in:

[0247] C risk : Collision risk value, calculated based on indicators such as close approach time (TCPA) and distance to closest approach (DCPA).

[0248] The original environmental variables are converted into standardized stress scores ranging from 0 to 1 using the f function mentioned above.

[0249] Among them, determine the collision risk value C risk , which is used to quantify the risk of a ship colliding with other ships during navigation.

[0250] Commonly used collision risk assessment indicators include:

[0251] Distance at Closest Point of Approach (DCPA): The closest distance between a ship and another ship is expected to reach in the track prediction.

[0252] TCPA (Time to Closest Point of Approach): The time required for a ship to reach DCPA with another ship.

[0253] Rate of Turn Angle (ROTA): The speed at which the other ship's position changes relative to your own ship.

[0254] Collision Risk Index (CRI) is a risk value that is evaluated based on multiple indicators.

[0255] Collision risk value C risk The calculation method includes the following process:

[0256] Step 1: Get other ship information

[0257] Obtain navigation data of other ships around you through the AIS system:

[0258] The course of his ship θ i , Speed ​​V i , position(x i ,y i ).

[0259] Step 2: Predict other ship movements

[0260] Calculate the motion parameters of other ships relative to your ship:

[0261] Relative speed

[0262] Where: V0, θ0: speed and heading of own ship.

[0263] Step 3: Calculate DCPA and TCPA

[0264] DCPA calculation:

[0265] TCPA Calculation:

[0266] in,

[0267] x0, y0: The horizontal and vertical coordinates of the current position of the ship.

[0268] x i ,y i : The horizontal and vertical coordinates of the current position of the i-th ship.

[0269] V rel,x : Relative velocity component of own ship relative to other ship in x direction.

[0270] Calculation formula: V rel,x =V i cosθ i -V0cosθ0.

[0271] V rel,y : Relative velocity component of own ship relative to other ship in y direction.

[0272] Calculation formula: V rel,y =V i sinθ i -V0sinθ0.

[0273] V0: own ship's speed.

[0274] θ0: The heading angle of own ship, calculated in clockwise direction relative to the true north direction.

[0275] V i : The speed of the i-th ship.

[0276] θ i : The heading angle of the i-th ship is calculated in a clockwise direction relative to the true north direction.

[0277] TCPA: Closest encounter time, that is, the time required for the ship and other ships to reach the closest encounter point.

[0278] DCPA: Distance of closest approach, that is, the minimum distance between own ship and other ships at the closest approach point.

[0279] : The square of the velocity components of the relative velocity in the x and y directions.

[0280] (x i -x0),(y i -y0): The position difference between the other ship and our ship in the x and y directions.

[0281] Among them, when TCPA<0, it means that the other ship has been missed and the collision risk is not considered.

[0282] Step 4: Calculate the collision risk value C risk

[0283] Use collision risk model: C risk =e -a·DCPA ·e -b·|TCPA| ,

[0284] in,

[0285] a, b: Model parameters that determine the impact of DCPA and TCPA on risk value, a, b>0.

[0286] Step 5: Accumulate the collision risk values ​​of multiple other ships

[0287] When there are multiple other ships around, consider the risks of each ship comprehensively:

[0288] Where, N: the number of other ships around; C risk,i : The collision risk value of the i-th ship.

[0289] C risk The calculation is based on the AIS data of other ships, using indicators such as TCPA and DCPA, and is calculated through a collision risk model.

[0290] For example, assuming the ship data:

[0291] Position (x0, y0) = (0, 0)

[0292] Heading θ0 = 0° (due north)

[0293] Speed ​​V0 = 10 knots

[0294] Other ship data (Other ship 1):

[0295] Position (x1, y1) = (1, -5) nautical miles

[0296] Heading θ1 = 90° (due east)

[0297] Speed ​​V1 = 12 knots

[0298] Model parameters:

[0299] a=0.5

[0300] b=0.1

[0301] Calculate the relative speed:

[0302] V rel,x =V1cosθ1-V0cosθ0=12cos90°-10cos0°=0-10=-10,

[0303] V rel,y=V1sinθ1-V0sinθ0=12sin90°-10sin0°=12-0=12,

[0304] Calculate TCPA:

[0305]

[0306] Calculate DCPA:

[0307]

[0308] Calculate the collision risk value C risk,1 :

[0309] C risk,1 =e -0.5×2.434 ·e -0.1×0.2877 ≈0.2885.

[0310] If there is only one other ship, then C risk =C risk,1 =0.2885.

[0311] C risk The calculation is based on the AIS data of other ships, using indicators such as TCPA and DCPA, and is calculated through a collision risk model.

[0312] Calculate each sub-pressure index, including using their respective weight coefficients w i , calculate the stress index of weather, sea conditions, navigation and traffic.

[0313] Calculate the comprehensive environmental pressure level S env : Using the environmental assessment model E, each sub-pressure index is divided into i Perform weighted summation to obtain the comprehensive environmental pressure level: S env =β1S weather +

[0314] β2S sea +β3S nav +β4S traffic .

[0315] Exemplarily, in a certain embodiment, it is assumed that the parameter weight coefficients are: β1=0.3, β2=0.2, β3=0.25, β4=0.25.

[0316] Weight coefficient of sub-pressure index:

[0317] Weather pressure index: w1=0.4, w2=0.3, w3=0.2, w4=0.1,

[0318] Sea pressure index: w5=0.6, w6=0.4,

[0319] Navigation pressure index: w7=0.7, w8=0.3,

[0320] Traffic pressure index: w9=0.5, w 10 =0.5,

[0321] First, you need to collect environmental data for the next time period, including:

[0322] Wind speed V w =15m / s, maximum wind speed

[0323] Visibility Vis = 5 km, Viss max =10km,Vis min =0km,

[0324] RainfallRain=5mm / h,Rain max =10mm / h,

[0325] Temperature Temp=35℃,Temp opt =25℃,Temp range =10℃,

[0326] Wave height H wave =3m,

[0327] Flow rate V c =1m / s,

[0328] Route complexity score R complexity =70,R max =100,

[0329] Estimated number of operations M frequency =5,M mdx =10,

[0330] Ship density D vessel =20,D max =30,

[0331] Collision risk value C rusk =0.6 (standardized).

[0332] In the next phase, a standardized stress score was calculated, including:

[0333]

[0334]

[0335]

[0336]

[0337]

[0338]

[0339]

[0340]

[0341]

[0342]

[0343] In the next stage, various sub-stress indices are calculated, including:

[0344] Weather Pressure Index S weather :S weather =0.4×0.75+0.3×0.5+0.2×0.5+0.1×1=0.65.

[0345] Sea pressure index S sea :S sea =0.6×0.6+0.4×0.5=0.36+0.2=0.56.

[0346] Navigation pressure index S nav :S nav =0.7×0.7+0.3×0.5=0.49+0.15=0.64.

[0347] Traffic Pressure Index S traffic :S traffic =0.5×0.6667+0.5×0.6=0.3333+0.3=0.6333.

[0348] Finally, calculate the comprehensive environmental pressure level S env :

[0349] S env =0.3×S weather +0.2×S sea +0.25×S nav +0.25×S traffic =0.6253.

[0350] Therefore, the comprehensive environmental pressure level S in the next period of time env =0.6253.

[0351] Obtain the ship's navigation environment data for the next time period through the AIS system and meteorological and sea condition service channels, process these data using the environmental assessment model E, and calculate the comprehensive stress level S of the crew's navigation environment env The model takes into account the influence of multiple environmental factors and transforms complex environmental data into a quantifiable stress index through standardization and weighted summation methods, providing an important basis for subsequent crew workload prediction and management.

[0352] In one embodiment, based on the comprehensive stress level of the sailing environment in which the crew is located in the next time period and the predicted value of the individual physiological load, the causal reasoning graph model C is input to predict the workload level of the crew in the future period, including:

[0353] Construct a causal reasoning graph model E to describe the causal relationship between variables.

[0354] The regression model was used to capture the effects of physiological load and environmental stress on workload, as well as their interaction.

[0355] Use parameter estimation methods to estimate model parameters.

[0356] In one embodiment, constructing the causal reasoning graph model C includes the following process:

[0357] First, the variable definitions:

[0358] : The predicted value of individual physiological load in the next time period t+Δt.

[0359] S env (t+Δt): The comprehensive pressure level of the navigation environment in the next time period t+Δt.

[0360] W(t+Δt): Crew workload level in the future (target variable).

[0361] Determine causal relationships, including defining individual physiological loads And the combined pressure of the navigation environment S env (t+Δt) has a direct impact on the crew workload level W(t+Δt).

[0362] In a certain embodiment, since there may be an interaction between individual physiological load and the comprehensive pressure of the navigation environment, the impact on workload is not a simple linear superposition. In order to capture the nonlinear effects of physiological load and environmental pressure on workload level, as well as the interaction between them, the following regression model is established:

[0363]

[0364] in:

[0365] W(t+Δt): crew workload level in the future.

[0366] f(·): A nonlinear function that represents the relationship between input variables and workload.

[0367] : The predicted value of individual physiological load in the next time period.

[0368] S env (t+Δt): The comprehensive pressure level of the navigation environment in the next time period.

[0369] θ: A set of model parameters, including weight coefficients and nonlinear exponents.

[0370] ∈: Error term, which represents the random error not explained by the model and is assumed to have a mean of zero.

[0371] In one embodiment, the regression model can be expressed as:

[0372]

[0373] in:

[0374] W(t+Δt): predicted value of workload level.

[0375] β0: Bias term, representing the basic workload level.

[0376] β1, β2, β3: Model coefficients, indicating the influence of the corresponding variables on the workload level.

[0377] :Individual physiological load prediction value.

[0378] S env (t+Δt): Comprehensive pressure level of navigation environment.

[0379] n1, n2, n3: nonlinear index, controlling the degree of nonlinearity of the variable.

[0380] In order to estimate the model parameters θ = {β0, β1, β2, β3, n1, n2, n3}, a nonlinear least square method can be used to minimize the error through an iterative optimization algorithm (such as the Levenberg Marquardt algorithm). In one embodiment, constructing a causal reasoning graph model C includes the following process:

[0381] Step 1: Obtain sufficient historical data, including:

[0382] Individual physiological load prediction

[0383] The comprehensive pressure level of the navigation environment Senv (t+Δt).

[0384] The corresponding crew workload level is W(t+Δt).

[0385] Clean, standardize or normalize the data to improve the stability and convergence of the model.

[0386] Step 2: Model training.

[0387] Set initial values ​​for the model parameters θ.

[0388] Select the loss function. The commonly used loss function is the mean square error (MSE):

[0389] in,

[0390] N: sample size.

[0391] W i : The actual workload level of the i-th sample.

[0392] : The workload level of the ith sample predicted by the model.

[0393] Use the optimization algorithm to minimize the loss function, update the parameter θ, and update the rule:

[0394] in:

[0395] θ (k) : parameter value for the kth iteration.

[0396] η: learning rate, controls the step size of parameter update.

[0397] : The gradient of the loss function with respect to the parameters θ.

[0398] Step 3: Divide the dataset into a training set and a validation set. Evaluate the prediction performance of the model on the validation set and calculate indicators such as MSE and root mean square error (RMSE). Adjust the model structure or retrain the model based on the performance of the model on the validation set.

[0399] In one embodiment, predicting the workload level includes:

[0400] Determine the trained model parameters θ = {β0, β1, β2, β3, n1, n2, n3}.

[0401] Individual physiological load prediction value for the next time period

[0402] The comprehensive pressure level of the navigation environment in the next period of time S env (t+Δt).

[0403] Calculate the workload level, including the following steps:

[0404] Step 1: Calculate each nonlinear term

[0405]

[0406]

[0407]

[0408] Step 2: Substitute into the model formula and calculate the workload prediction value W(t+Δt)=β0+β1h1+β2h2+β3h3.

[0409] Exemplarily, assume that the model parameters are: β0=-0.5, β1=0.8, β2=0.6, β3=0.4, n1=1.5, n2=2.0, n3=1.0.

[0410] Input variables: (Normalization, for example, dividing by the maximum value 30, gives S env (t+Δt)=0.6253.

[0411] Calculation of workload levels includes:

[0412] Calculating nonlinear terms

[0413] h1=(0.75) 1.5 =0.6495,

[0414] h2=(0.6253) 2.0 =0.3909,

[0415] h3=(0.75×0.6253) 1.0 =0.4690;

[0416] Calculate workload forecast:

[0417] W(t+Δt)=-0.5+0.8×0.6495+0.6×0.3909+0.4×0.4690=0.4417.

[0418] That is, the predicted workload level is 0.4417.

[0419] In one embodiment, the initially predicted workload level is first revised based on the estimated rest time of the next time period obtained from the work and rest schedule.

[0420] Based on the environmental factors of the route and the personnel situation in the work and rest arrangements for the next time period, a second revision is made to the workload level after the first revision.

[0421] Record the intermediate value and the value before correction: record the intermediate value between two correction processes and the workload level before each correction to obtain the multi-stage, multi-factor load and the relationship between the corrected load and the cause.

[0422] The above relationship and data are fed back to the control center for decision support and management optimization. In one embodiment, obtaining the work / rest arrangement of personnel in the next time period includes obtaining the work and rest arrangement of all crew members in the next time period from the work and rest schedule, and calculating the personnel availability coefficient and determining the individual rest time:

[0423] For the personnel availability factor P avail :

[0424]

[0425] in,

[0426] P avail : Personnel availability coefficient, which indicates the proportion of personnel working in the next time period.

[0427] N working : Number of crew members working in the next time period.

[0428] N total : Total number of crew members on board.

[0429] Individual rest time R:

[0430] R i : The estimated rest time of the i-th crew member in the next time period (unit: hours).

[0431] The impact of individual rest time on personal workload is reflected in the fact that crew members who do not have enough rest time may have increased fatigue and an increased workload level W; crew members who have sufficient rest time may have a reduced workload level W.

[0432] The impact of the personnel availability coefficient on individual workload is reflected in the fact that when the personnel availability coefficient is low, there are fewer people actually working and the workload of each person increases.

[0433] The impact of environmental factors on the overall workload is reflected in the environmental pressure S env When it is higher, more workload is required to meet environmental challenges, and insufficient staffing can exacerbate individual workloads.

[0434] In one embodiment, based on the estimated rest time of the next time period obtained from the work and rest schedule, as well as the environmental factors of the route and the personnel situation in the work and rest schedule of the next time period, a correction is performed to obtain W. adjusted include:

[0435] W adjusted =W×F(R)×G(P avail ,S env ),

[0436] in:

[0437] W adjusted : Corrected individual workload level.

[0438] W: Uncorrected individual workload level (calculated by the above model).

[0439] F(R): Correction factor based on individual rest time.

[0440] G(P avail ,S env ): Correction factor based on personnel availability factor and environmental pressure.

[0441] In one embodiment, the correction factor F(R) can be expressed as:

[0442] F(R): Individual rest time correction factor.

[0443] λ1: Rest time influence coefficient, reflecting the impact of rest time on workload, λ1>0.

[0444] R: The estimated rest time of the individual in the next time period (unit: hours).

[0445] R opt : Ideal optimal rest time (unit: hours).

[0446] Among them, when R <R opt When F(R)>1, it means that the rest is insufficient and the workload level increases;

[0447] When R>R opt When F(R)<1, it means that the rest is sufficient and the workload level is reduced.

[0448] In one embodiment, the correction factor G(P avail ,S env ) can be expressed as:

[0449] G(P avail ,S env): Comprehensive correction factor, taking into account the effects of personnel availability and environmental pressure.

[0450] λ2: influence coefficient of personnel availability and environmental pressure, λ2>0.

[0451] P avail : Personnel availability coefficient, 0 <P avail ≤1.

[0452] P opt : Ideal staff availability coefficient, usually 1.

[0453] S env : Comprehensive environmental pressure level, 0≤S env ≤1.

[0454] μ: Environmental pressure impact index, μ≥0.

[0455] Among them, when P avail <P opt When G(P avail ,S env )>1, indicating insufficient staff and increased workload level; when P avail >P opt When G(P avail ,S env )<1, indicating that there is a surplus of personnel and the workload level is reduced; environmental pressure S env The higher the correction factor, the greater the impact of the correction factor on the workload level. In one embodiment, based on the estimated rest time for the next time period obtained from the work and rest schedule, as well as based on the environmental factors of the route and the personnel situation in the work and rest schedule for the next time period, W is obtained. adjusted include:

[0456]

[0457] in:

[0458] W adjusted : Corrected individual workload level.

[0459] W: Uncorrected individual workload level.

[0460] R: The individual's expected rest time in the next time period.

[0461] R opt : Ideal optimal rest time.

[0462] λ1: Rest time influence coefficient.

[0463] P avail : Personnel availability coefficient.

[0464] Popt : Ideal staff availability factor.

[0465] λ2: Personnel availability and environmental pressure influence coefficient.

[0466] S env : Comprehensive environmental stress level.

[0467] μ: Environmental pressure impact index.

[0468] The model takes into account the nonlinear relationship of multiple influencing factors and can predict the workload level more accurately.

[0469] For example, it is known that the uncorrected individual workload level is: W = 0.5; the individual rest time is: R = 1.5 hours; the ideal optimal rest time is: R opt =2 hours; Rest time impact coefficient: λ1=0.3; Personnel availability coefficient: Ideal personnel availability factor: P opt =1; Human and environmental impact coefficient: λ2=0.5; Comprehensive environmental pressure level: S env =0.7;

[0470] Environmental pressure impact index: μ=2.

[0471] Correction to get W adjusted include:

[0472] Calculate F(R):

[0473] Calculate G(P avail ,S env ): Calculate the corrected workload level: W adjusted =W×F(R)×G(P avail ,S env )=0.5655.

[0474] The above process shows that since the rest time is lower than the ideal rest time, the correction factor F(R)>1, and the workload level increases by about 7.79%.

[0475] Personnel and environmental corrections: Personnel availability is lower than ideal and environmental stress is high, resulting in a correction factor G(P avail ,S env )>1, the workload level increases by about 5.02%.

[0476] The corrected workload level increased by approximately 13.1% over the uncorrected level.

[0477] In one embodiment, recording the intermediate values ​​between two correction processes and the workload level before each correction includes:

[0478] Record the various phases of workload levels, including:

[0479] The uncorrected individual workload level W calculated from the model, record variable: W initial =W.

[0480] The first revised workload level W adjusted1 ,include:

[0481] Based on the correction factor F(R) of the individual rest time, calculate the first corrected workload level: W adjusted1 =W initial ×F(R).

[0482] Record variables: F(R): individual rest time correction factor; W adjusted1 : Workload level after the first revision.

[0483] The second revised workload level W adjusted , including the correction factor G(P avail ,S env ), calculate the second revised workload level: W asjusted =W adjusted1 ×G(P avail ,S env ).

[0484] Record variable: G(P avail ,S env ): comprehensive correction factor, W adiusted : Final corrected workload level.

[0485] In one embodiment, the multi-stage, multi-factor load and modified load and cause are fed back to the control center, including:

[0486] For each correction stage, record: the workload level value, the value of the correction factor, the correction increment and the percentage impact, and a detailed description of the reason for the correction.

[0487] For example, create a data record table as follows:

[0488]

[0489] The present invention utilizes the individual physiological load model P and builds a model according to the physiological data of each crew member, thereby achieving accurate assessment of the workload of different individuals and enhancing the individual adaptability and prediction accuracy of the model.

[0490] Based on current and historical physiological load data, the present invention dynamically predicts the physiological load of an individual in a future period of time through a time series prediction model to ensure the real-time and forward-looking nature of the prediction results.

[0491] The present invention introduces an environmental assessment model E to systematically assess the comprehensive pressure level of the navigation environment in the next time period, quantitatively incorporates environmental factors into workload prediction, and improves the sensitivity and responsiveness of the model to changes in the external environment.

[0492] The present invention organically combines the predicted value of individual physiological load with the environmental stress level through the causal reasoning graph model C, deeply explores the causal relationship between various factors, and improves the interpretability and accuracy of workload prediction.

[0493] The present invention comprehensively utilizes the synergy of multi-source data and multiple models to achieve adaptive prediction of crew workload. It can dynamically adjust the prediction results according to real-time data and environmental changes to ensure that the crew workload is always within a controllable range under various complex navigation conditions.

[0494] The present invention collects and integrates the crew's physiological characteristic data, historical physiological load parameters and navigation environment data to comprehensively and multi-dimensionally reflect the crew's working status, thereby improving the information content and comprehensive analysis capabilities of the prediction model.

[0495] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0496] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0497] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0498] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0499] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware, wherein the name of a unit does not, in some cases, limit the unit itself.

[0500] The above introduces the preferred embodiments of the present invention, which is intended to make the spirit of the present invention clearer and easier to understand, but is not intended to limit the present invention. All modifications, substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection outlined by the claims attached to the present invention.

Claims

1. A crew workload adaptive prediction method based on multi-source heterogeneous data fusion, comprising: Collecting physiological characteristic data of the crew members, inputting it into the individual physiological load model P, and obtaining the current physiological load of the individual; Based on the current physiological load of the individual and the physiological load parameters at historical moments, inputting into a time series prediction model to obtain a predicted value of the individual physiological load in the next time period; Obtain the navigation environment data of the crew's ship in the next time period; Based on the navigation environment data of the next time period, input the environmental assessment model E to obtain the comprehensive stress level of the navigation environment in which the crew is located in the next time period; Based on the comprehensive stress level of the sailing environment in which the crew members are located in the next time period and the predicted value of the individual physiological load in the next time period, the causal reasoning graph model C is input to predict the workload level of the crew members.

2. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 1 is characterized in that: Wearable devices are used to collect physiological characteristics data of crew members in real time, including: Heart rate HR(t), heart rate variability HRV(t), activity measured by accelerometer Activity(t), skin temperature Temp(t), skin conductivity EDA(t).

3. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 2 is characterized in that: Define the individual physiological load model P, which is used to calculate the physiological load L(t) of the crew at the current time t; the model considers the nonlinear accumulation of fatigue and the physiological dynamic recovery process, and the formula is as follows: in, L(t): physiological load level at the current time t; L(t-Δt): physiological load level at the previous moment t-Δt; HR(t): heart rate measured at the current time t; HR rest : individual's resting heart rate; HRV(t): heart rate variability measured at the current time t; HRV rest : Individual resting heart rate variability; Activity(t): the amount of activity at the current time t; Δt: time interval; α1, α2, α3: individualized weight coefficients; n1, n2, n3: nonlinear index; γ: Physiological load recovery rate parameter.

4. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 3 is characterized in that: Based on the individual's current physiological load and physiological load parameters at historical moments, the individual's physiological load in the next time period is predicted, including: When predicting the physiological load value of the next time period, the current and historical physiological load data, as well as the model parameters, are used to predict the physiological load of the time period (t+Δt) 5. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 1 is characterized in that: The navigation environment data of the ship where the crew is located in the next time period is obtained through the AIS system or the meteorological and sea condition service channel, and based on this, the environmental assessment model E is input to obtain the comprehensive stress level of the navigation environment in which the crew is located in the next time period.

6. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 5 is characterized in that: Define an environmental assessment model E, which is used to convert the above-obtained navigation environment data into the comprehensive stress level S of the navigation environment in which the crew is located. env : S env =E(environmental parameter)=β1S weather +β2S sea +β3S nav +β4S traffic ; in, S env : Comprehensive environmental stress level; Weight coefficient β i : represents the weight of each environmental factor on the comprehensive pressure, satisfying β1+β2+β3+β4=1, and β i ≥0; Among them, the environmental pressure sub-indicators include: Weather Pressure Index S weather :S weather =w1f wind (V w )+w2f visibibity (Vis)+w3f rain (Rain)+w4f temp (Temp); Sea pressure index S sea :S sea =w5f wave (H wave )+w6f current (V c ); Navigation pressure index S nav :S nav =w7f route (R complexity )+w8f maneuver (M frequency );Traffic Pressure Index S traffic :S traffic =w9f vesseldensity (D vessel )+w 10 f collisionrisk (C risk ); in, Weight coefficient w i : represents the influence weight of each specific factor on the corresponding pressure index, satisfying that the sum of the weight coefficients in the corresponding pressure index is 1, and w i ≥0; Function f converts environmental variables into standardized stress scores, ranging from 0 to 1.

7. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 1 is characterized in that: Based on the comprehensive stress level of the crew's navigation environment in the next time period and the predicted value of individual physiological load, the causal reasoning graph model C is input to predict the crew's workload level in the future, including: Construct a causal reasoning graph model C to describe the causal relationship between variables; Based on regression model, we capture the effects of physiological load and environmental stress on workload, as well as their interaction; Use parameter estimation methods to estimate model parameters.

8. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 7 is characterized in that: Constructing the causal reasoning graph model C includes the following processes: First, the variable definitions: The predicted value of individual physiological load in the next time period t+Δt; S env (t+Δt): the comprehensive pressure level of the navigation environment in the next time period t+Δt; W(t+Δt): crew workload level in the future; Determine causal relationships, including defining individual physiological loads and the combined pressure of the navigation environment S env (t+Δt) has a direct impact on the crew workload level W(t+Δt).

9. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 7, characterized in that: The regression model can be expressed as: in: W(t+Δt): predicted value of workload level; β0: bias term, indicating the basic workload level; β1, β2, β3: model coefficients, indicating the influence of the corresponding variables on the workload level; Individual physiological load prediction value; S env (t+Δt): comprehensive pressure level of navigation environment; n1, n2, n3: nonlinear index, controlling the nonlinearity of the variable; In order to estimate the model parameters θ = {β0, β1, β2, β3, n1, n2, n3}, a nonlinear least squares method is used to minimize the error through an iterative optimization algorithm.

10. The crew workload adaptive prediction method based on multi-source heterogeneous data fusion according to claim 1, characterized in that: Make a first correction to the initially predicted workload level based on the estimated rest time for the next period obtained from the work-rest schedule; Based on the environmental factors of the route and the personnel situation in the work and rest arrangement of the next period, the workload level after the first revision is revised for the second time; The intermediate values ​​between two correction processes and the workload level before each correction are recorded to obtain the multi-stage, multi-factor load and the relationship between the corrected load and the cause and feed it back to the control center.

Images