Ship fire risk prediction method and device based on multi-algorithm fusion, and medium

Abstract

The invention discloses a ship fire risk prediction method and device based on multi-algorithm fusion and a medium. The method comprises the following steps: acquiring ship sensor data; respectively processing the ship sensor data by using a multi-sensor data fusion algorithm and a support vector machine classification algorithm; a ship comprehensive danger index and / or a preliminary alarm sign are / is output through a multi-sensor data fusion algorithm, and a ship state label and / or a classification alarm sign are / is output through a support vector machine classification algorithm; a gray fuzzy algorithm is combined with the obtained ship sensor data, the output of the multi-sensor data fusion algorithm and the output of the support vector machine classification algorithm to carry out ship fire risk prediction, and a prediction alarm mark is output; and when it is detected that at least one of the preliminary alarm sign, the classified alarm sign and the predicted alarm sign exists, ship fire early warning operation is carried out. Early warning can be realized at the initial stage of a fire, fire spreading is effectively controlled, and fire hazards are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship safety, in particular to a ship fire risk prediction method, device and medium based on multi-algorithm fusion. BACKGROUND

[0002] During the navigation of a ship, fire is always a key hidden danger that poses a major threat to the safety of the ship and the lives and property of personnel. Due to the relatively closed space of the ship and the fact that it often carries flammable substances such as fuel and cargo, if a fire accident occurs and is not controlled in time, the fire can easily spread rapidly with the help of the ventilation system or the cabin structure, ultimately leading to disastrous consequences. This risk is particularly pronounced during ocean voyages, as external rescue forces are usually difficult to reach quickly, so it is crucial to strengthen fire prevention and early intervention measures.

[0003] Currently, fire alarm systems and flammable gas detection systems have been widely used on ships. These devices mainly detect based on indicators such as smoke, heat, or flammable gas concentration, but only trigger an alarm when a fire has formed or developed to a certain extent.

[0004] However, these devices cannot effectively detect early risk signs before a fire occurs, such as overheating of electrical equipment, aging of wiring, or accumulation of flammable materials, thus missing the best prevention opportunity and making subsequent remedial measures ineffective due to the spread of the fire. SUMMARY

[0005] Therefore, the embodiments of the present application provide a ship fire risk prediction method, device and medium based on multi-algorithm fusion.

[0006] The first aspect of the present application provides a ship fire risk prediction method based on multi-algorithm fusion, comprising the following steps: Obtaining ship sensor data; Using a multi-sensor data fusion algorithm and a support vector machine classification algorithm to process the ship sensor data respectively; outputting a ship comprehensive risk index and / or a preliminary alarm flag through the multi-sensor data fusion algorithm, and outputting a ship state label and / or a classification alarm flag through the support vector machine classification algorithm; Using a gray fuzzy algorithm to combine the obtained ship sensor data, the output of the multi-sensor data fusion algorithm, and the output of the support vector machine classification algorithm to perform ship fire risk prediction and output a prediction alarm flag; When at least one of the preliminary alarm flag / classification alarm flag / prediction alarm flag is detected, a ship fire warning operation is performed.

[0007] Further, the ship sensor data includes environmental data collected by at least one of a smoke detector, a temperature detector, and a flame detector distributed in each area of the ship. After the step of acquiring the ship sensor data, the method further includes a step of preprocessing the ship sensor data; and the preprocessing specifically includes normalization processing.

[0008] Further, the processing of the ship sensor data using the multi-sensor data fusion algorithm specifically includes the following steps: detecting the validity of the ship sensor data and eliminating invalid ship sensor data; setting a risk weight for each area of the ship, and assigning a sensor weight to a ship sensor in each area according to the risk weight of each area of the ship; weighting the ship sensor data combined with the weight to obtain a ship comprehensive risk index and output the ship comprehensive risk index; comparing the obtained ship comprehensive risk index with a preset fusion alarm threshold value, and outputting a preliminary alarm flag when the ship comprehensive risk index is higher than the fusion alarm threshold value.

[0009] Further, the processing of the ship sensor data using the support vector machine classification algorithm specifically includes the following steps: converting the sensor data into a fire feature vector; inputting the fire feature vector into a support vector machine model that has been trained, and the support vector machine model calculates the probability of the fire feature vector belonging to each state category through a decision function; the state categories include a category representing a normal state of the ship and abnormal state categories representing a gas leakage state, a smoldering state, and an open fire state; calculating the sum of the probabilities of the abnormal state categories, and outputting a classification alarm flag when the sum of the probabilities of the abnormal state categories is higher than a preset abnormal state threshold value.

[0010] Further, the support vector machine model is trained through the following steps: acquiring ship sensor historical data; the ship sensor historical data includes ship sensor data when the ship is in a normal state, a gas leakage state, a smoldering state, and an open fire state; after converting the ship sensor historical data into a fire training feature vector, manually labeling the fire training feature vector; constructing a support vector machine with a radial basis function as a kernel function, and using the support vector machine to solve an optimal hyperplane to maximize the interval between fire training feature vectors of different state categories; determining the best model parameters of the support vector machine through cross-validation, and using the support vector machine with the best model parameters as the support vector machine model that has been trained.

[0011] Further, the ship fire risk prediction using the gray fuzzy algorithm specifically comprises the following steps: Gray prediction is performed on the ship sensor data to obtain a ship sensor data prediction value and a development coefficient at a future time; The ship sensor data prediction value at the future time, the development coefficient, the ship comprehensive risk index and the ship state label are taken as input variables for fuzzy processing to obtain multiple fuzzy inputs; A fire prediction fuzzy rule is established, and the fuzzy inputs are substituted into the fire prediction fuzzy rule to obtain a fuzzy fire risk; The centroid of the fuzzy fire risk is calculated, and the centroid value is taken as a comprehensive fire risk value; When the comprehensive fire risk value is higher than a preset fire risk threshold, a prediction alarm flag is output.

[0012] Further, the gray prediction on the ship sensor data specifically comprises the following steps: The ship sensor data at multiple times is arranged into a time sequence as an original feature sequence; The original feature sequence is accumulated to obtain an exponential feature sequence; The original feature sequence and the exponential feature sequence are substituted into a gray differential equation for solving to obtain a development coefficient and a gray action amount; The development coefficient and the gray action amount are substituted into a time response function to obtain an accumulated value at a future time; The accumulated value at the future time is subjected to a cumulative reduction operation to obtain a ship sensor data prediction value at the future time.

[0013] Further, the input variables are subjected to fuzzy processing, specifically comprising the following steps: For each input variable, a set of fuzzy sets is defined; each fuzzy set corresponds to a language value for describing the state of the variable and covers the entire value range of the input variable; A membership function is designed, and for the current value of each input variable, the membership degree of each fuzzy set is calculated; The membership degree values of the multiple fuzzy sets corresponding to each input variable are taken as fuzzy inputs.

[0014] Another aspect of the present application discloses an electronic device comprising a processor and a memory; The memory is used to store a program; The processor executes the program to realize the above-mentioned ship fire risk prediction method based on multi-algorithm fusion.

[0015] In another aspect of the present application, a computer readable storage medium is provided, and the storage medium stores a program, and the program is executed by a processor to implement the ship fire risk prediction method based on multi-algorithm fusion.

[0016] The present application also discloses a computer program product or a computer program, which comprises computer instructions stored in a computer readable storage medium. A processor of a computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to enable the computer device to perform the foregoing method.

[0017] The present application has the following beneficial effects: the ship fire risk prediction method based on multi-algorithm fusion, device and medium of the present application fuse multi-sensor data fusion algorithm, SVM classification algorithm and grey fuzzy algorithm to perform early prediction of ship fire. The present application calculates the comprehensive risk index and the state label of the ship through the multi-sensor data fusion algorithm and the SVM classification algorithm executed in parallel, respectively, and directly performs fire warning when the two calculated values exceed the threshold value. Then, the grey fuzzy algorithm is used to process the sensor data of the ship, the comprehensive risk index and the state label of the ship to obtain the comprehensive fire risk value of the ship for early fire risk warning. The present application effectively detects various early signs of fire through multi-algorithm fusion, can realize fire warning in the early stage of fire, effectively controls the spread of fire, reduces fire hazards, and ensures the safe, reliable and stable operation of the ship.

[0018] Additional aspects and advantages of the present application will be described in the following description part, some of which will become apparent from the following description, or will be understood through practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0020] Figure 1 is a basic step flowchart of the ship fire risk prediction method based on multi-algorithm fusion of the present application; Figure 2 is a step flowchart of the multi-sensor data fusion algorithm executed in the present application; Figure 3 is a step flowchart of the SVM classification algorithm executed in the present application; Figure 4 is a step flowchart of the grey fuzzy algorithm executed in the present application; Figure 5 This is a schematic diagram of the structure of an electronic device according to the present invention; Figure 6 This is a schematic diagram of a computer-readable storage medium structure according to the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0022] The first embodiment of the present invention provides a method for predicting ship fire risk based on multi-algorithm fusion, such as... Figure 1 As shown, it includes the following steps: S1. Acquire ship sensor data; S2. Process ship sensor data using multi-sensor data fusion algorithm and support vector machine classification algorithm respectively; output ship comprehensive hazard index and / or preliminary alarm sign through multi-sensor data fusion algorithm, and output ship status label and / or classification alarm sign through support vector machine classification algorithm; S3. Using a gray fuzzy algorithm combined with the acquired ship sensor data, the output of a multi-sensor data fusion algorithm, and the output of a support vector machine classification algorithm, ship fire risk is predicted, and a predicted alarm sign is output. S4. When at least one of the preliminary alarm sign, classification alarm sign, or prediction alarm sign is detected, a ship fire early warning operation shall be performed.

[0023] This invention provides an early prediction of ship fires by integrating multi-sensor data fusion algorithms, SVM classification algorithms, and gray fuzzy algorithms.

[0024] The implementation process of the algorithm for each step of this invention is explained in detail below: S1. Acquire ship sensor data.

[0025] This invention, through the rational deployment of various sensors such as smoke detectors, heat detectors, flame detectors, and combustible gas sensors in key areas of a ship, including the engine room, cargo hold, and crew living quarters, in accordance with regulatory requirements, enables the collection of environmental data in these critical areas. For example, in the engine room, a combination of uniform distribution and focused monitoring can be used, with smoke detectors and heat detectors installed at predetermined intervals (e.g., 5-8 meters) to form dual detection coverage. Simultaneously, combustible gas sensors are densely deployed at critical locations where combustible gases are prone to accumulate or leak, such as the fuel pump room, fuel storage area, and ventilation duct inlets and outlets. Furthermore, the density and type of smoke, heat, and flame detectors can be adjusted according to the type of cargo in the cargo hold (e.g., cargo that may smolder) and the characteristics of living facilities in the living quarters.

[0026] In some embodiments, the sensors installed on the ship are not limited to smoke detectors, heat detectors, flame detectors, and combustible gas sensors; other types of ship sensors can also be installed for more diversified fire early warning.

[0027] After the ship's sensors are deployed, data (including smoke concentration, temperature, flame intensity, and combustible gas concentration) are collected periodically and connected to the central processing unit via redundant cables or wireless communication modules. At the ship's central processing unit, multi-sensor data fusion algorithms, SVM state classification algorithms, and gray-fuzzy prediction algorithms are executed to achieve early prediction of ship fires.

[0028] In some embodiments, after acquiring the ship sensor data, a preprocessing step is included. Since different sensor data may have different dimensions—for example, temperature data may range from 0 to 150°C, while smoke concentration data may only range from 0 to 5 obs / m—these different dimensions can lead to significant deviations in the algorithm's calculations. Therefore, this embodiment of the invention normalizes the acquired ship sensor data using the following formula: S_norm=S / Smax, where Smax is the smoke concentration alarm threshold; T_norm = (T - T_ambient) / (T_alarm - T_ambient), where T_ambient is the ambient temperature and T_alarm is the temperature rise alarm threshold; F_norm=F, where F is the flame intensity, which is a relative value between 0 and 1; G_norm = G / G_alarm, where G_alarm is the combustible gas alarm threshold.

[0029] By normalizing ship sensor data, the contribution of each type of sensor data to fire prediction is made more balanced, enabling more accurate early fire prediction.

[0030] S2. Process ship sensor data using a multi-sensor data fusion algorithm and a support vector machine classification algorithm respectively; output the ship's comprehensive hazard index and / or preliminary alarm signs through the multi-sensor data fusion algorithm, and output the ship's status label and / or classification alarm signs through the support vector machine classification algorithm.

[0031] like Figure 2 As shown, in step S2, the ship's sensor data is processed using a multi-sensor data fusion algorithm, specifically including the following steps: S2-A1. Detect the validity of ship sensor data and remove invalid ship sensor data.

[0032] Multi-sensor data fusion algorithms are used to comprehensively analyze and process data collected by different sensors. In this embodiment of the invention, before applying the multi-sensor data fusion algorithm, the validity of the ship's sensor data is first verified, for example, to determine whether there is data loss or obvious anomalies, or to check whether the data is within a reasonable range. Invalid ship sensor data may be generated due to the failure of a single sensor. By removing these invalid data, the accuracy of the output results of the multi-sensor data fusion algorithm can be improved.

[0033] S2-A2. Set the risk weight for each area of ​​the ship, and assign sensor weights to the ship's sensors in each area according to the risk weight of each area.

[0034] In this embodiment of the invention, since different areas of a ship have different fire risk characteristics, it is necessary to assign different weights to different types of ship sensors according to the ship area where the sensor is located, while ensuring that the weight coefficients of each area of ​​the ship satisfy the condition that the sum is 1.

[0035] For example, in areas with high temperatures, dense equipment, and high fire risk, such as the engine room, the weights of temperature sensors and flame detectors will be set higher. In this embodiment of the invention, the sensor weights for the engine room are set as follows: The weight of the temperature sensor is W_T = 0.5; The weight of the smoke detector is W_S=0.3; The flame detector weight W_F = 0.15; The weight of the combustible gas sensor is W_G=0.05.

[0036] For areas prone to smoke, such as cargo holds, the data weight of smoke detectors is increased accordingly. In this embodiment of the invention, the following sensor weights are set for the cargo hold: The weight of the smoke detector is W_S=0.6; The weight of the temperature sensor is W_T = 0.3; The flame detector weight W_F = 0.1; The weight of the combustible gas sensor is W_G=0.0.

[0037] Other areas, such as fuel storage areas and crew living quarters, are also assigned different sensor weights accordingly; for example, the fuel storage area is assigned a higher weight to combustible gas sensors, while the crew living quarters are assigned a higher weight to smoke detectors and flame detectors.

[0038] S2-A3. The ship sensor data is weighted and calculated to obtain the ship's comprehensive risk index and output it.

[0039] In this embodiment of the invention, the comprehensive ship hazard index for different shipping areas is calculated separately, specifically using the following formula: Danger_Index = (S_norm * W_S) + (T_norm * W_T) + (F_norm * W_F) + (G_norm * W_G).

[0040] For example, if at a certain moment the sensor data in the engine room are S_norm=0.4, T_norm=0.7, F_norm=0.1, and G_norm=0.3 (normalized value), then the ship's comprehensive risk index calculated based on this set of data is: Danger_Index = (0.4 * 0.3) + (0.7 * 0.5) + (0.1 * 0.15) + (0.3 *0.05) = 0.27.

[0041] In this embodiment of the invention, the comprehensive ship hazard index for each of the different areas of the ship is calculated, and finally a set of comprehensive ship hazard indices for that moment is obtained.

[0042] S2-A4. The overall ship risk index is compared with the preset fusion alarm threshold. When the overall ship risk index is higher than the fusion alarm threshold, a preliminary alarm sign is output.

[0043] In this embodiment of the invention, the fusion alarm threshold is set to 0.7. When the calculated ship comprehensive hazard index shows that at least one region's ship comprehensive hazard index is higher than the fusion alarm threshold, a preliminary alarm flag is generated and output to achieve early warning for ship fire prediction.

[0044] In this embodiment of the invention, a multi-sensor data fusion algorithm is used to achieve complementary information from multiple sources of ship sensors, effectively filtering environmental noise and minimizing the possibility of false alarms caused by single sensor failure or interference. A weighted average fusion calculation is used to derive a comprehensive ship hazard index for different regions, making the algorithm's judgment more consistent with the actual situation in each region. The multi-sensor data fusion algorithm enables the perception of weak correlations and anomalies among multiple parameters, providing a robust early warning method for ship fires.

[0045] like Figure 3 As shown, in step S2, the support vector machine classification algorithm is used to process the ship sensor data, specifically including the following steps: S2-B1. Convert sensor data into fire feature vectors.

[0046] Support Vector Machine (SVM) classification algorithm is a supervised classification model used for classifying and identifying the states of fire and combustible gas leaks. Based on the classification function obtained through training, the SVM model transforms the system state identification problem into a pattern classification problem. It can distinguish between various complex states such as "normal", "early stage of fire (smoldering)", "open flame stage of fire" and "combustible gas leak", and accurately determine which preset category the current state of the ship belongs to, thus providing a reliable basis for fire early warning of ships.

[0047] In this embodiment of the invention, statistical features are first extracted from the time-series data collected by ship sensors to form a fire feature vector. Specifically, this embodiment uses a sliding event window to process ship sensor data, performs statistical analysis on the sensor data within the window, extracts statistical features, and combines them into a fire feature vector.

[0048] For example, statistical characteristics may include the following features:

[0049] The combined fire feature vector can be represented in the following form: X_current = [Mean_S, Mean_T, Mean_F, Mean_G, Slope_S, Slope_T, Slope_F, Slope_G, Max_S, Max_T, Max_F, Max_G, Var_S, Var_T].

[0050] S2-B2. Input the fire feature vector into the trained support vector machine model. The support vector machine model calculates the probability that the fire feature vector belongs to each state category through the decision function. The state categories include categories that represent the ship in a normal state and abnormal state categories that represent the ship in a gas leak state, smoldering state, and open flame state.

[0051] The Support Vector Machine (SVM) used in steps S2-B2 is capable of handling non-linear classification problems. The SVM classifier finds an optimal hyperplane by mapping fire feature vectors to a high-dimensional space, which maximizes the margin between samples of different categories (such as "normal" and "open flame").

[0052] As a specific example, the support vector machine model is trained through the following steps: S2-B21. Acquire historical data of ship sensors; the historical data of ship sensors includes data of ship sensors when the ship is in normal condition, gas leak condition, smoldering condition, and open flame condition.

[0053] This step involves acquiring ship sensor data under various conditions from sources such as historical sensor records during actual ship operation, ship fire experiment data, and fire test databases. This data serves as the ship's historical sensor data. The acquired data includes complete data from four types of sensors: smoke concentration, temperature, flame intensity, and combustible gas concentration, and the sampling frequency of these data is adjusted to be consistent.

[0054] S2-B22. After converting historical data from ship sensors into fire training feature vectors, manually label the fire training feature vectors.

[0055] After acquiring ship sensor data, the historical sensor data is converted into fire feature vectors using the same statistical analysis method as in step S2-B1. Then, the fire feature vectors are manually labeled to give them accurate ship status labels (e.g., Label 0-3 represent normal status, smoldering status, open flame status, and gas leak status, respectively).

[0056] S2-B23. Construct a support vector machine with radial basis functions as kernel functions, and use the support vector machine to solve for the optimal hyperplane, so as to maximize the margin between the fire training feature vectors of different state categories.

[0057] Since fire data is typically non-linearly separable, this embodiment of the invention selects the radial basis function (RBF) kernel, which handles complex non-linear classification problems, as the kernel function of the SVM. Its formula is as follows: ; in γ It can control the width of the kernel, thus affecting the model complexity.

[0058] The training process for SVM is as follows: Labeled ship sensor data is input into the SVM model, and the SVM model finds the optimal hyperplane that maximizes the margin between samples of different classes by solving the following optimization problem: ; Its constraints are: ; in, w It is the hyperplane normal vector. b It is a bias term. C This is a penalty parameter used to control the tolerance for misclassification. It is a slack variable.

[0059] S2-B24. Determine the optimal model parameters of the support vector machine through cross-validation, and use the support vector machine with the optimal model parameters as the support vector machine model for training.

[0060] By adjusting hyperparameters C and γ This allows for the acquisition of multiple sets of model parameters, which are then used to determine the optimal parameters for the SVM model using cross-validation. In this embodiment, k-fold cross-validation is employed. K-fold cross-validation involves randomly dividing the training set into k subsets of similar size; then performing k rounds of training and validation: each round uses k-1 subsets of data as the training set, and the remaining subset as the validation set. Finally, the average of the k rounds of validation results is taken as the performance metric for this parameter combination.

[0061] After determining the optimal model parameters, the support vector machine under the optimal model parameters is used as the support vector machine model to complete the training, and the fire feature vector X_current is classified in real time. The category to which the fire feature vector belongs to each state is obtained (e.g., Label 0=0.7, which means that the probability of the ship being in a normal state is 70%).

[0062] S2-B3. Calculate the total probability of abnormal state categories. When the total probability of abnormal state categories is higher than the preset abnormal state threshold, output a classification alarm flag.

[0063] In this embodiment of the invention, the abnormal state category refers to any state category other than the normal state category Label 0, including smoldering, open flame, and gas leak. This embodiment sets the abnormal state threshold to 0.7. When the sum of the probabilities of the abnormal state categories exceeds the preset abnormal state threshold, a classification alarm flag is directly output to achieve early warning for ship fire prediction.

[0064] This invention uses an SVM classification algorithm to identify the current state of a ship. By training, it learns the complex correlation patterns between features such as smoke concentration and temperature change rate in a smoldering state, enabling the ship's central processing unit to identify various early stages of a fire, thus gaining valuable time for firefighting and evacuation.

[0065] S3. Using a gray fuzzy algorithm combined with the acquired ship sensor data, the output of a multi-sensor data fusion algorithm, and the output of a support vector machine classification algorithm, ship fire risk is predicted, and a predicted alarm sign is output.

[0066] like Figure 4 As shown, in step S3, the grey fuzzy algorithm is used to predict the risk of ship fires, which specifically includes the following steps: S3-1. Perform grey prediction on ship sensor data to obtain the predicted values ​​and development coefficients of ship sensor data at future times; The grey algorithm makes short-term predictions of ship fire trends based on sensor data. By predicting the trends of real-time collected data such as temperature, smoke concentration, and combustible gas concentration, it can anticipate the changes in these parameters in advance.

[0067] As a specific example, gray prediction of ship sensor data includes the following steps: S3-1-1. Organize the ship sensor data from multiple time points into a time series as the original feature sequence.

[0068] In this embodiment of the invention, ship sensor data from multiple time points are combined to form an original feature sequence.

[0069] Taking temperature-sensing data as an example, the combined original feature sequence is as follows: ; in, This represents the temperature value at the k-th time point, where n is the sequence length.

[0070] S3-1-2. Accumulate the original feature sequences to obtain the exponential feature sequences; Accumulating the original feature sequences can make the feature sequences exhibit a dominant pattern. Taking temperature sensing data as an example, the accumulation of the original feature sequences is accomplished using the following formula: ; The accumulated exponential feature sequence is shown below: .

[0071] S3-1-3. Substitute the original feature sequence and the exponential feature sequence into the grey differential equation to solve for the development coefficient and grey action quantity.

[0072] The basic structure of the grey differential equation is shown below: ; in, a The development coefficient reflects the rate of increase in the fire trend. bThis is a grey effect quantity, reflecting the impact of external intervention.

[0073] In this embodiment of the invention, after discretizing the grey differential equation, the exponential feature sequence is substituted into the equation to obtain... a and b The specific value.

[0074] S3-1-4. Substitute the development coefficient and the gray action quantity into the time response function to obtain the cumulative value at future time moments; The time response function has the following form: ; in This represents the cumulative temperature value at the next time point.

[0075] S3-1-5. Perform a cumulative subtraction operation on the accumulated value at future time points to obtain the predicted value of the ship sensor data at future time points.

[0076] The predicted temperature value for the next time point is restored through the cumulative subtraction operation: .

[0077] The same calculations are performed on sensor data such as smoke concentration, flame intensity, and gas concentration to obtain the predicted parameter values ​​for future moments.

[0078] S3-2. The predicted values ​​of ship sensor data at future times, the development coefficient, the ship comprehensive risk index, and the ship status label are used as input variables and fuzzified to obtain multiple fuzzy inputs; In step S3-2, the input variables are fuzzified, which specifically includes the following steps: S3-2-1. For each input variable, define a set of fuzzy sets; each fuzzy set corresponds to a linguistic value, which is used to describe the state of the variable and covers the entire value range of the input variable; As a specific example, the fuzzy set defined for each variable is shown in the table below:

[0079] The linguistic values ​​of each fuzzy set cover the entire range of values ​​for the input variable, so there will be no situation where the input variable does not belong to any fuzzy set.

[0080] S3-2-2. Design a membership function to calculate the membership degree of each input variable to each fuzzy set for the current value of the input variable. Membership functions are used to calculate the degree to which an exact value belongs to each fuzzy set, with values ​​between 0 and 1. Membership functions include trigonometric functions, trapezoidal functions, or Gaussian functions, etc. This embodiment of the invention uses a trigonometric function to calculate membership.

[0081] The membership function of a triangle is expressed in the following form: ; Where x represents the input variable, c represents the center point of the triangle, and w represents the half-width of the triangle.

[0082] Taking temperature prediction as an example, the fuzzy set parameters for temperature prediction are: Low temperature: center point c=20℃, half width w=20℃, function coverage range is 0°C-40°C.

[0083] Medium temperature: center point c=50℃, half width w=20℃, function coverage range is 30°C-70°C.

[0084] High temperature: center point c=80℃, half width w=20℃, function coverage range is 60°C-100°C.

[0085] For a predicted temperature input variable of 40℃, its membership degree is calculated using the following steps: Membership degree belonging to "low temperature": ; Membership degree belonging to "medium temperature": ; Membership degree belonging to "high temperature": ; Therefore, the input variable belongs to the "medium temperature" fuzzy set, with a membership degree of 0.5.

[0086] The membership degrees of the remaining input variables are also calculated using the same process.

[0087] S3-2-3. Use the membership values ​​of the multiple fuzzy sets corresponding to each input variable as fuzzification inputs.

[0088] The output of fuzzification is the membership value of each input variable for each fuzzy set; these values ​​will serve as the input for fuzzy inference. The output can be represented as: Predicted temperatures: {Low temperature: 0, Medium temperature: 0.5, High temperature: 0}; Temperature change trend: {Slow rise: 0, Moderate rise: 0.5, Rapid rise: 0.5}; Risk level: {Low risk: 0, Medium risk: 0.25, High risk: 0.75}.

[0089] ... These membership values ​​are used in the fuzzy inference steps of S3-3.

[0090] S3-3. Establish fuzzy rules for fire prediction, and substitute the fuzzy inputs into the fuzzy rules for fire prediction to obtain fuzzy fire risk; This invention establishes an "IF-THEN" rule as a fuzzy rule for fire prediction based on expert experience (fire protection regulations). For example, the fuzzy rule for fire prediction is as follows: Rule 1: If the predicted temperature is high and the temperature trend is rapidly increasing, then the fire risk is extremely high.

[0091] Rule 2: If the predicted temperature is moderate and the hazard index is high, then the fire risk is high.

[0092] Rule 3: If the SVM classification status is smoldering and the smoke concentration prediction is high, then the fire risk is moderate.

[0093] Rule 4: IF all inputs are normal, THEN fire risk is low.

[0094] The above rules are just examples of fuzzy rules for fire prediction. In practice, there are dozens of fuzzy rules for fire prediction that can cover various fuzzy inference inputs.

[0095] After establishing fuzzy rules for fire prediction, this embodiment of the invention substitutes the fuzzy inference input into all relevant fuzzy rules for calculation, and calculates the strength of the conclusion based on the membership degree. For example, for each rule, the matching degree of the preconditions is calculated; then, based on the matching degree, the strength of the conclusion is generated; finally, the outputs of all triggered rules are aggregated to form a comprehensive output fuzzy set.

[0096] S3-4. Calculate the centroid of the fuzzy fire risk and use the centroid value as the comprehensive fire risk value; After obtaining the output fuzzy set, this embodiment of the invention performs defuzzification processing on the fuzzy set to obtain the comprehensive fire risk value.

[0097] Taking the centroid method as an example, the centroid of the output fuzzy set can represent the comprehensive fire risk value. The centroid is calculated using the following formula: ; in, It is the membership function of the output fuzzy set, and x is the range of risk values.

[0098] The comprehensive fire risk value, calculated using the above formula, represents the overall risk level of the input variables.

[0099] S3-5. When the overall fire risk value is higher than the preset fire risk threshold, output a predictive alarm sign.

[0100] In steps S3-5, the comprehensive fire risk value obtained through defuzzification is compared with a preset fire risk threshold. In this embodiment of the invention, the fire risk threshold is set to 0.8. If the risk value exceeds the threshold, it is determined that there is a high fire hazard.

[0101] In this embodiment of the invention, a gray fuzzy algorithm is used to predict the trends of real-time collected data such as temperature, smoke concentration, and combustible gas concentration, allowing for advance forecasting of these parameters' changes. Furthermore, pre-defined fuzzy rules (such as a high probability of fire due to rapid temperature increases and high smoke concentration) enable fuzzy reasoning regarding the likelihood of a fire. By combining gray and fuzzy algorithms, potential hazards can be warned in advance, and the hazard level can be more accurately determined when a dangerous situation occurs.

[0102] S4. When at least one of the preliminary alarm sign, classification alarm sign, or prediction alarm sign is detected, a ship fire early warning operation shall be performed.

[0103] In step S4, when at least one preliminary alarm sign / classification alarm sign / predictive alarm sign is detected, it indicates that one of the aforementioned multi-sensor data fusion algorithm, SVM classification algorithm, and gray fuzzy algorithm has concluded that there is a fire risk. At this point, immediately control the ship to cut off the power to the ventilation fans to stop ventilation and prevent the fire from spreading due to airflow; cut off the power to the oil pumps to stop fuel supply and prevent the fire from escalating due to continued fuel combustion; close the fireproof dampers and electric louvers to form a fireproof isolation zone and prevent the fire from spreading, etc., to ensure the safe, reliable, and stable operation of the ship.

[0104] In this embodiment of the invention, the collaborative operation of multi-sensor data fusion algorithm, SVM classification algorithm and gray fuzzy algorithm can effectively detect various early signs of fire, realize fire early warning in the early stage of fire, effectively control the spread of fire, reduce fire hazards, and ensure the safe, reliable and stable operation of ships.

[0105] Figure 5This is a schematic diagram of the electronic device proposed in the second embodiment of the present invention. In this embodiment, the memory stores program instructions for implementing the ship fire risk prediction method based on multi-algorithm fusion in any of the above embodiments. The processor executes the program instructions stored in the memory to perform ship fire risk prediction based on multi-algorithm fusion. The processor can also be called a CPU (Central Processing Unit). The processor may be an integrated circuit chip with signal processing capabilities. The processor can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0106] The methods described in the first embodiment of the present invention are applicable to the embodiments of the present electronic device. The specific functions implemented by the embodiments of the present electronic device are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0107] Figure 6 This is a schematic diagram of the structure of a computer-readable storage medium according to the third embodiment of the present invention. The computer-readable storage medium of the fourth embodiment of the present invention stores program instructions capable of implementing the above-described ship fire risk prediction method based on multi-algorithm fusion. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned computer-readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0108] The methods described in the first embodiment of the present invention are applicable to the computer-readable storage medium embodiment. The specific functions implemented by the computer-readable storage medium embodiment are the same as those in the above method embodiment, and the beneficial effects achieved are also the same as those achieved by the above method.

[0109] This embodiment also provides a computer program product. When the computer program product is run on a computer, it causes the computer to perform the above-mentioned related steps to realize the ship fire risk prediction method based on multi-algorithm fusion provided in the above embodiment.

[0110] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of the present invention are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0111] Those skilled in the art will understand that modules in the device of the embodiments of the present invention can be adaptively modified and placed in one or more devices different from those embodiments. Modules, units, or components in the embodiments of the present invention can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the corresponding claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the corresponding claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0112] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0113] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0114] Furthermore, the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. In particular, for embodiments such as apparatus and devices, since they are basically similar to the method embodiments, the relevant parts can be referred to the description of the method embodiments. The apparatus, devices, and other embodiments described above are merely illustrative, and the modules, units, etc., described as separate components may or may not be physically separate, that is, they may be located in one place or distributed in multiple places, such as nodes in a system network. Specifically, some or all of the modules and units can be selected according to actual needs to achieve the purpose of the above-described embodiment solutions. Those skilled in the art can understand and implement this without creative effort.

[0115] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0116] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0117] Furthermore, the terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying relative importance, or implicitly specifying the number of technical features indicated in this embodiment. Therefore, features defined with terms such as "first" and "second" in the embodiments of this invention can explicitly or implicitly indicate that the embodiment includes at least one of those features. In the description of this invention, the word "multiple" means at least two or more, such as two, three, four, etc., unless otherwise explicitly specified in the embodiments.

[0118] In embodiments of the present invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, components, features, and elements with the same names in different embodiments of the present invention may have the same meaning or different meanings, the specific meaning of which must be determined by its interpretation in that specific embodiment or further in conjunction with the context of that specific embodiment.

[0119] Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention. Other embodiments of the present invention will readily conceive of by considering the specification and practicing the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

Claims

1. A method for predicting ship fire risk based on multi-algorithm fusion, characterized in that, Includes the following steps: Acquire ship sensor data; The ship's sensor data is processed using a multi-sensor data fusion algorithm and a support vector machine classification algorithm, respectively. The multi-sensor data fusion algorithm outputs a comprehensive ship hazard index and / or a preliminary alarm sign, while the support vector machine classification algorithm outputs a ship status label and / or a classification alarm sign. The grey fuzzy algorithm is used in combination with the acquired ship sensor data, the output of the multi-sensor data fusion algorithm and the output of the support vector machine classification algorithm to predict ship fire risk and output a predicted alarm sign. When at least one of the preliminary alarm sign, classification alarm sign, or prediction alarm sign is detected, a ship fire early warning operation is initiated.

2. The ship fire risk prediction method based on multi-algorithm fusion according to claim 1, characterized in that, The ship sensor data includes environmental data collected by at least one of smoke detectors, heat detectors, and flame detectors distributed in various areas of the ship. Following the step of acquiring ship sensor data, a step of preprocessing the ship sensor data is also included; the preprocessing specifically includes normalization processing.

3. The ship fire risk prediction method based on multi-algorithm fusion according to claim 1, characterized in that, The process of processing the ship's sensor data using a multi-sensor data fusion algorithm specifically includes the following steps: Detect the validity of ship sensor data and eliminate invalid ship sensor data; Set risk weights for each area of ​​the ship, and assign sensor weights to the ship's sensors in each area based on the risk weights for each area. The ship sensor data is weighted and calculated to obtain the ship's comprehensive risk index, which is then output. The ship's overall risk index is compared with the preset fusion alarm threshold. When the ship's overall risk index is higher than the fusion alarm threshold, a preliminary alarm sign is output.

4. The ship fire risk prediction method based on multi-algorithm fusion according to claim 1, characterized in that, The process of using a support vector machine classification algorithm to process the ship sensor data specifically includes the following steps: Convert sensor data into fire feature vectors; The fire feature vector is input into the trained support vector machine model, and the support vector machine model calculates the probability that the fire feature vector belongs to each state category through a decision function; the state categories include categories that represent the ship in a normal state and abnormal state categories that represent the ship in a gas leak state, smoldering state, and open flame state. Calculate the sum of probabilities for each abnormal state category. When the sum of probabilities for each abnormal state category exceeds a preset abnormal state threshold, output a classification alarm flag.

5. The ship fire risk prediction method based on multi-algorithm fusion according to claim 4, characterized in that, The support vector machine model is trained through the following steps: Acquire historical data from ship sensors; the historical data from ship sensors includes data from ship sensors when the ship is in normal condition, gas leak condition, smoldering condition, and open flame condition; After converting historical data from ship sensors into fire training feature vectors, the fire training feature vectors are manually labeled. Construct a support vector machine with radial basis functions as kernel functions, and use the support vector machine to solve for the optimal hyperplane, so as to maximize the margin between the fire training feature vectors of different state categories; The optimal model parameters of the support vector machine are determined by cross-validation, and the support vector machine with the optimal model parameters is used as the support vector machine model for training.

6. The ship fire risk prediction method based on multi-algorithm fusion according to claim 1, characterized in that, The method of using a gray fuzzy algorithm for ship fire risk prediction specifically includes the following steps: Grey prediction is performed on ship sensor data to obtain the predicted values ​​and development coefficients of ship sensor data at future times; The predicted values ​​of ship sensor data at the future time, the development coefficient, the ship comprehensive risk index, and the ship status label are used as input variables and fuzzified to obtain multiple fuzzified inputs. Establish fuzzy rules for fire prediction, and substitute the fuzzified input into the fuzzy rules for fire prediction to obtain fuzzy fire risk; Calculate the centroid of the fuzzy fire risk and use the centroid value as the comprehensive fire risk value; When the overall fire risk value is higher than the preset fire risk threshold, a predictive alarm flag is output.

7. The ship fire risk prediction method based on multi-algorithm fusion according to claim 6, characterized in that, The gray prediction of ship sensor data specifically includes the following steps: Ship sensor data from multiple time points are compiled into a time series, which serves as the original feature sequence. The original feature sequences are accumulated to obtain the exponential feature sequences; Substitute the original feature sequence and the exponential feature sequence into the grey differential equation to solve for the development coefficient and grey action quantity. Substituting the development coefficient and gray action quantity into the time response function, the cumulative value at future time moments is obtained; The accumulated values ​​at future times are subtracted to obtain the predicted values ​​of ship sensor data at future times.

8. The ship fire risk prediction method based on multi-algorithm fusion according to claim 6, characterized in that, The input variables are fuzzified, specifically including the following steps: For each input variable, a set of fuzzy sets is defined; each fuzzy set corresponds to a linguistic value, which is used to describe the state of the variable and covers the entire value range of the input variable. Design a membership function to calculate the membership degree of each input variable to each fuzzy set for the current value of the input variable. The membership values ​​of the multiple fuzzy sets corresponding to each input variable are used as fuzzification inputs.

9. An electronic device, characterized in that, Including the processor and memory; The memory is used to store programs; The processor executes the program to implement a ship fire risk prediction method based on multi-algorithm fusion as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The storage medium stores a program, which is executed by a processor to implement a ship fire risk prediction method based on multi-algorithm fusion as described in any one of claims 1-8.

Images