A method for predicting and fusion of data information of hazardous chemicals transportation vehicles

Abstract

A method for predicting and fusing data information of hazardous chemical transport vehicles collects state information through sensors, and then sends the collected data into the improved seagull algorithm optimization neural network to predict the state parameters at the next moment and send them into the fuzzy fusion algorithm to fuse the prediction results, thereby predicting the operating state of the hazardous chemical transport vehicle at the next moment. It can perform real-time and predictive alarms for single sensors, and can also perform real-time and predictive alarms through multi-sensor data fusion; and send its results to the vehicle-mounted terminal for display. The present invention can effectively improve the safety of hazardous chemical transport vehicles during the transportation process and reduce accident losses.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent prediction of hazardous chemicals logistics, and in particular to a method for predicting and fusing data information of hazardous chemicals transport vehicles. Background Art

[0002] Dangerous chemicals refer to items that have explosive, flammable, toxic, corrosive, radioactive and other properties, which can easily cause personal injury and property damage during transportation, loading and unloading, and storage and custody, and therefore require special protection. They are referred to as hazardous chemicals.

[0003] With the rapid development of my country's economy and chemical industry, hazardous chemicals (hereinafter referred to as hazardous chemicals) as important chemical raw materials are used more and more in various fields, and the demand and transportation volume of hazardous chemicals are increasing year by year. As the transportation volume of hazardous chemicals increases year by year, traffic safety and smoothness have become the top priority, especially the storage and logistics of hazardous chemicals. Due to its flammable, explosive, poisonous and polluting characteristics, accidents during transportation will cause huge losses.

[0004] Transportation safety management directly involves the safety of life and property of the general public, especially the safety management of road dangerous goods transportation. The existing management of hazardous chemical transportation cannot effectively and comprehensively monitor the transportation of hazardous chemicals, resulting in certain risks at the source and during the transportation of hazardous chemicals. Summary of the invention

[0005] In response to the above-mentioned technical problems, the present technical scheme provides a method for predicting and fusing data information of hazardous chemicals transport vehicles. It combines sensors to collect relevant data of hazardous chemicals transport vehicles, and sends the collected data to the improved Seagull algorithm to optimize the neural network. It predicts the state parameters at the next moment and sends them to the fuzzy fusion algorithm to fuse the predicted results, thereby predicting the operating state of the hazardous chemicals transport vehicle at the next moment, and sending the results to the on-board terminal for display, which can effectively improve the safety of hazardous chemicals transport vehicles during the transportation process and reduce accident losses; it can effectively solve the above-mentioned problems.

[0006] The present invention is achieved through the following technical solutions:

[0007] A method for predicting and fusing data information of hazardous chemicals transport vehicles collects state information through sensors, and then sends the collected data to the improved seagull algorithm optimized neural network to predict the state parameters of the next moment and send them to the fuzzy fusion algorithm to fuse the prediction results, so as to predict the operation state of the hazardous chemicals transport vehicle at the next moment. It can perform real-time and predictive alarms for a single sensor, and can also perform real-time and predictive alarms through multi-sensor data fusion; finally, the prediction results are output and sent to the vehicle terminal for display; the specific operation steps include:

[0008] Step 1: Use various sensors to collect status information of hazardous chemicals transport vehicles;

[0009] Step 2: Upload the collected sensor data to the improved Seagull algorithm optimized neural network for data processing;

[0010] Step 3: Get multiple prediction parameters, and send the prediction parameters to the fuzzy fusion algorithm to perform data fusion on the prediction parameters;

[0011] Step 4: Obtain the fusion result of the prediction data and predict the operation status of the hazardous chemicals transportation vehicle at the next moment;

[0012] Step 5: Send the predicted results to the vehicle terminal for display.

[0013] Furthermore, the various sensors described in step 1 include an acceleration sensor, a vehicle speed sensor, and a posture sensor installed on the vehicle body, and a gas concentration sensor and a temperature and humidity sensor installed on or inside the tank body. The acceleration sensor, vehicle speed sensor, posture sensor, gas concentration sensor and temperature and humidity sensor of the tank body are connected to the sensor group module signal of the microprocessor through a signal line or a wireless module; the microprocessor is also connected to a positioning module and an alarm module.

[0014] Furthermore, the improved seagull algorithm optimization neural network described in step 2 is a seagull algorithm that integrates the grey wolf hierarchy, levy flight strategy and t distribution variation to improve the seagull optimization algorithm;

[0015] At the beginning of the iteration, according to the hierarchy of the gray wolf algorithm strategy, some candidate solutions are generated around the seagulls with better fitness near their positions, which can increase the diversity of the population and improve the local exploration ability of the seagull optimization algorithm;

[0016] In the process of iterative optimization, the introduction of Levy flight strategy can improve the global search ability and convergence speed of the Seagull algorithm. After the introduction of Levy flight strategy, the Seagull algorithm is:

[0017]

[0018] Where i is the current iteration number, The individual position of the seagull flying in Levy is introduced for the current position of the seagull, P s (i) is the current position of the seagull, P bs (i) indicates the current optimal seagull position;

[0019]

[0020] Where r1 and r2 are random numbers between [0, 1] that follow a standard normal distribution, Levy (λ) is the random step length, λ is the power, and β = 1.5 here;

[0021]

[0022] In the later stage of iteration, as the seagulls tend to the optimal position as a whole, if the individuals in the optimal position are located near the local extreme value of the problem, the solution will fall into the local extreme value. Therefore, the t-distribution variation is introduced to perturb the optimal individual position of the seagulls, which helps to jump out of the local extreme value and improve the global optimization ability of the seagull optimization algorithm.

[0023]

[0024] in, represents the individual position of the seagull after the position of the current optimal seagull is disturbed by the t distribution, t(itre) represents the t distribution based on the number of iterations itre as the degree of freedom, P bs (i) indicates the current optimal seagull position;

[0025] When f t <f x When , it means that the new position obtained after the t-distribution mutation disturbance is better than the fitness of the current best seagull. The individual with better fitness is retained as the best seagull and added to the seagull group, and the seagull with the worst fitness is eliminated. t Represents the fitness value of the new position, f x Represents the original fitness value.

[0026] Furthermore, the specific operation steps of the improved Seagull algorithm are as follows:

[0027] Step 1: Initialize algorithm parameters, including number of iterations, population size, problem dimension, and optimization range;

[0028] Step 2: Initialize the individual positions of seagulls and calculate the fitness value of each seagull;

[0029] step3: start iteration;

[0030] Step 4: Calculate the optimal individual fitness of the seagull population and record the position of the optimal individual;

[0031] Step 5: Generate candidate solutions based on the gray wolf hierarchy around the seagulls with better fitness, and calculate the fitness value of the corresponding solution;

[0032] Step 6: Generate random numbers, perform spiral flight and Levy flight migration behaviors of seagulls based on the random number value and the probability of 0.5, and calculate the fitness value of each individual seagull;

[0033] Step 7: Update the location information and fitness value of the optimal seagull;

[0034] Step 8: Perturb the position of the optimal seagull according to the t distribution formula, compare the fitness of the mutated individual with that of the optimal seagull, retain the individual with better fitness as the optimal seagull, add it to the seagull group, and eliminate the seagull with the worst fitness;

[0035] Step 9: When the number of iterations does not exceed the maximum number of iterations, repeat step 4 to step 8, otherwise the iteration loop ends;

[0036] Step 10: Output the optimal seagull position and fitness value.

[0037] Furthermore, the weights and thresholds of the neural network are optimized by the improved Seagull algorithm, and the BP neural network is optimized by LGSOA to improve the prediction performance of the neural network. The main steps are:

[0038] Step a: Initialize the parameters, the number of seagulls, and encode the weights and thresholds of the BP neural network to be optimized as the initial seagull population;

[0039] Step b: Initialize the position of the seagull and process its position using a formula to prevent position overlap;

[0040] Step c: Calculate the fitness of all current seagulls and find the best one as the best seagull in this iteration. The single output fitness is calculated using the root mean square error, and the multi-output fitness is the Euclidean distance error.

[0041] Step d: Update the position of each seagull according to the formula;

[0042] Step e: Determine whether the number of iterations has been reached, if yes, output the final position as the optimal seagull position, otherwise return to step c;

[0043] Step f: Decode the optimal output into the initial weights and thresholds of the BP neural network, and train the neural network until the requirements are met.

[0044] Furthermore, after the LGSOA is used to optimize the BP neural network, a prediction alarm is performed for the single sensor data, and the obtained data is sent to the fuzzy fusion algorithm to fuse the prediction results, thereby predicting the operating status of the hazardous chemicals transport vehicle at the next moment.

[0045] Beneficial Effects

[0046] Compared with the prior art, the method for predicting and fusing data information of hazardous chemicals transport vehicles proposed in the present invention has the following beneficial effects:

[0047] (1) This technical solution combines sensors to collect relevant data of hazardous chemicals transport vehicles, and sends the collected data to the improved Seagull algorithm to optimize the neural network, predict the state parameters at the next moment and send them to the fuzzy fusion algorithm to fuse the prediction results, thereby predicting the operating state of the hazardous chemicals transport vehicle at the next moment, and sending the results to the on-board terminal for display, which can effectively improve the safety of hazardous chemicals transport vehicles during the transportation process and reduce accident losses.

[0048] (2) This technical solution proposes an improved seagull algorithm, which integrates the gray wolf hierarchy, levy flight strategy and t distribution variation to improve the seagull optimization algorithm. At the beginning of the iteration, according to the hierarchy of the gray wolf algorithm strategy, some candidate solutions are generated around the seagulls with better fitness near their positions, which can increase the diversity of the population and improve the local exploration ability of the seagull optimization algorithm. In the process of iterative optimization, the Levy flight strategy is introduced to realize the migration of seagull individuals, which can improve the global search ability and convergence speed of the seagull algorithm. In the later stage of the iteration, since the seagulls tend to the optimal position as a whole, if the individuals in the optimal position are located near the local extreme value of the problem, the solution will fall into the local extreme value. Therefore, the introduction of t distribution variation and the perturbation of the optimal individual position of the seagull can help to jump out of the local extreme value and improve the global optimization ability of the seagull optimization algorithm.

[0049] (3) This technical solution uses an improved Seagull algorithm to optimize the BP neural network and improve the prediction performance of the neural network. In the improved Seagull algorithm to optimize the neural network, the state parameters at the next moment are predicted and sent to the fuzzy fusion algorithm to fuse the prediction results, thereby predicting the operating state of the hazardous chemicals transportation vehicle at the next moment. Real-time and predictive alarms can be performed for a single sensor, and real-time and predictive alarms can also be performed through multi-sensor data fusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a specific schematic diagram of the present invention.

[0051] Figure 2It is a schematic diagram of the on-board terminal framework of a hazardous chemicals transport vehicle in the present invention.

[0052] Figure 3 (a), (b), and (c) are comparison diagrams of BP, SOA-BP, and LGSOA-BP training curves, respectively.

[0053] Figure 4 (a), (b), and (c) are comparison diagrams of the predicted values ​​and true values ​​of the velocity, acceleration, and posture training sets, respectively.

[0054] Figure 5 (a), (b), and (c) are the speed, acceleration, and posture training error analysis diagrams respectively.

[0055] Figure 6 (a), (b), and (c) are comparison diagrams of the predicted values ​​and the actual values ​​of the velocity, acceleration, and posture test sets, respectively.

[0056] Figure 7 (a), (b), and (c) are error analysis diagrams of the velocity, acceleration, and posture test sets, respectively.

[0057] Figure 8 This is a fitness curve comparison chart. DETAILED DESCRIPTION

[0058] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Under the premise of not departing from the design concept of the present invention, various modifications and improvements made by ordinary persons in the art to the technical solutions of the present invention should all fall within the protection scope of the present invention.

[0059] Embodiment 1:

[0060] like Figure 1 As shown, a method for predicting and fusing data information of a hazardous chemical transport vehicle is provided. The hazardous chemical transport vehicle is provided with a vehicle terminal, and the vehicle terminal includes various sensors installed at designated positions of the vehicle, a wireless transmission module, a display module and a voice playback module.

[0061] The various sensors include acceleration sensors, vehicle speed sensors, attitude sensors and Beidou satellite positioning modules installed on the vehicle body for collecting vehicle information; gas concentration sensors, temperature and humidity sensors and pressure sensors for collecting hazardous chemicals information. The acceleration sensors, vehicle speed sensors, attitude sensors, Beidou satellite positioning modules, gas concentration sensors, temperature and humidity sensors and pressure sensors are connected to the sensor group module of the microprocessor through signal lines or wireless modules; the microprocessor is also connected to a positioning module and an alarm module.

[0062] The main functions of the vehicle terminal system are: receiving and processing sensor output data, receiving and processing positioning module data, and wireless transmission of vehicle terminal data through the wireless transmission module. If there is an abnormal situation, the driver will be notified in time.

[0063] The data information prediction and fusion method of hazardous chemicals transportation vehicles mainly collects state information through sensors, and then sends the collected data to the improved seagull algorithm to optimize the neural network, predicts the state parameters at the next moment and sends them to the fuzzy fusion algorithm to fuse the prediction results, so as to predict the operation state of the hazardous chemicals transportation vehicle at the next moment. It can perform real-time and prediction alarms for single sensors, and can also perform real-time and prediction alarms through multi-sensor data fusion; finally, the prediction results are output and sent to the vehicle terminal for display; the specific operation steps include:

[0064] Step 1: Use various sensors to collect status information of hazardous chemicals transport vehicles;

[0065] Step 2: Upload the collected sensor data to the improved Seagull algorithm optimized neural network for data processing;

[0066] The improved seagull algorithm optimization neural network is a seagull algorithm that integrates the gray wolf hierarchy, levy flight strategy and t distribution variation to improve the seagull optimization algorithm;

[0067] At the beginning of the iteration, according to the hierarchy of the gray wolf algorithm strategy, some candidate solutions are generated around the seagulls with better fitness near their positions, which can increase the diversity of the population and improve the local exploration ability of the seagull optimization algorithm;

[0068] In the process of iterative optimization, the introduction of Levy flight strategy can improve the global search ability and convergence speed of the Seagull algorithm. After the introduction of Levy flight strategy, the Seagull algorithm is:

[0069]

[0070] Where i is the current iteration number, The individual position of the seagull flying in Levy is introduced for the current position of the seagull, P s (i) is the current position of the seagull, P bs (i) indicates the current optimal seagull position;

[0071]

[0072] Where r1 and r2 are random numbers between [0, 1] that follow a standard normal distribution, Levy (λ) is the random step length, λ is the power, and β = 1.5 here;

[0073]

[0074] In the later stage of iteration, as the seagulls tend to the optimal position as a whole, if the individuals in the optimal position are located near the local extreme value of the problem, the solution will fall into the local extreme value. Therefore, the t-distribution variation is introduced to perturb the optimal individual position of the seagulls, which helps to jump out of the local extreme value and improve the global optimization ability of the seagull optimization algorithm.

[0075]

[0076] in, represents the individual position of the seagull after the position of the current optimal seagull is disturbed by the t distribution, t(itre) represents the t distribution based on the number of iterations itre as the degree of freedom, P bs (i) indicates the current optimal seagull position;

[0077] When f t <f x When , it means that the new position obtained after the t-distribution mutation disturbance is better than the fitness of the current best seagull. The individual with better fitness is retained as the best seagull and added to the seagull group, and the seagull with the worst fitness is eliminated. t Represents the fitness value of the new position, f x Represents the original fitness value.

[0078] The specific operation steps of the improved Seagull algorithm are:

[0079] Step 1: Initialize algorithm parameters, including number of iterations, population size, problem dimension, and optimization range;

[0080] Step 2: Initialize the individual positions of seagulls and calculate the fitness value of each seagull;

[0081] step3: start iteration;

[0082] Step 4: Calculate the optimal individual fitness of the seagull population and record the position of the optimal individual;

[0083] Step 5: Generate candidate solutions based on the gray wolf hierarchy around the seagulls with better fitness, and calculate the fitness value of the corresponding solution;

[0084] Step 6: Generate random numbers, perform spiral flight and Levy flight migration behaviors of seagulls based on the random number value and the probability of 0.5, and calculate the fitness value of each individual seagull;

[0085] Step 7: Update the location information and fitness value of the optimal seagull;

[0086] Step 8: Perturb the position of the optimal seagull according to the t distribution formula, compare the fitness of the mutated individual with that of the optimal seagull, retain the individual with better fitness as the optimal seagull, add it to the seagull group, and eliminate the seagull with the worst fitness;

[0087] Step 9: When the number of iterations does not exceed the maximum number of iterations, repeat step 4 to step 8, otherwise the iteration loop ends;

[0088] Step 10: Output the optimal seagull position and fitness value.

[0089] Step 3: Get multiple prediction parameters, and send the prediction parameters to the fuzzy fusion algorithm to perform data fusion on the prediction parameters;

[0090] The weights and thresholds of the neural network are optimized by using the improved Seagull algorithm, and the BP neural network is optimized using LGSOA to improve the prediction performance of the neural network. The main steps are:

[0091] Step a: Initialize the parameters, the number of seagulls, and encode the weights and thresholds of the BP neural network to be optimized as the initial seagull population;

[0092] Step b: Initialize the position of the seagull and process its position using a formula to prevent position overlap;

[0093] Step c: Calculate the fitness of all current seagulls and find the best one as the best seagull in this iteration. The single output fitness is calculated using the root mean square error, see formula (4); the multi-output fitness is the Euclidean distance error, see formula (5);

[0094]

[0095] Step d: Update the position of each seagull according to the formula;

[0096] Step e: Determine whether the number of iterations has been reached, if yes, output the final position as the optimal seagull position, otherwise return to step c;

[0097] Step f: Decode the optimal output into the initial weights and thresholds of the BP neural network, and train the neural network until the requirements are met; the BP neural network uses LGSOA to optimize the BP neural network; after the BP neural network is optimized using the LGSOA, predict and alarm the single sensor data, and send the obtained data to the fuzzy fusion algorithm to fuse the prediction results.

[0098] Step 4: Obtain the fusion result of the prediction data and predict the operation status of the hazardous chemicals transportation vehicle at the next moment;

[0099] like Figure 3-Figure 8 As shown, taking the condition of a hazardous chemical vehicle as an example, the parameters that need to be measured are speed, acceleration, and posture.

[0100] Use 8 groups in Tables 1 to 5 for training and 3 groups for testing. The previous data predicts the next data.

[0101] Table 1. Raw data

[0102]

[0103] Table 2 Comparison of the actual value of the training set and the three predicted values

[0104]

[0105] Table 3 Comparison of training set prediction error values

[0106]

[0107] Table 4 Comparison of the true value of the test set and the three predicted values

[0108]

[0109] Table 5 Comparison of prediction error values ​​of test set

[0110]

[0111] Its MAE stands for mean absolute error, RMSE stands for root mean square error, MSE stands for mean square error, MAPE stands for mean absolute percentage error, and R2 stands for goodness of fit. The larger the R2, the better, and the maximum is 1. The smaller the other error indicators, the better.

[0112] Figure 3 It indicates that the improved SOA optimized BP training has the fastest convergence speed and the smallest error.

[0113] Figure 4-7 It represents the comparison of the predicted values ​​and error values ​​of each parameter.

[0114] Figure 8 It represents the comparison of fitness curves between improved SOA optimized BP and SOA optimized BP.

[0115] Step 5: Send the predicted results to the vehicle terminal for display.

[0116] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for predicting and fusing data information of hazardous chemicals transportation vehicles, characterized by: The state information is collected through sensors, and the collected data is then sent to the improved seagull algorithm to optimize the neural network, predict the state parameters at the next moment and send them to the fuzzy fusion algorithm to fuse the prediction results, so as to predict the operation state of the hazardous chemicals transportation vehicle at the next moment. It can perform real-time and predictive alarms for single sensors, and can also perform real-time and predictive alarms through multi-sensor data fusion; finally, the prediction results are output and sent to the vehicle terminal for display; the specific operation steps include: Step 1: Use various sensors to collect status information of hazardous chemicals transport vehicles; Step 2: Upload the collected sensor data to the improved seagull algorithm optimization neural network for data processing; the improved seagull algorithm optimization neural network is a seagull algorithm that integrates the gray wolf hierarchy, levy flight strategy and t distribution variation to improve the seagull optimization algorithm; At the beginning of the iteration, according to the hierarchy of the gray wolf algorithm strategy, some candidate solutions are generated around the seagulls with better fitness near their positions, which can increase the diversity of the population and improve the local exploration ability of the seagull optimization algorithm; In the process of iterative optimization, the introduction of Levy flight strategy can improve the global search ability and convergence speed of the Seagull algorithm. After the introduction of Levy flight strategy, the Seagull algorithm is: Where i is the current iteration number, The individual position of the seagull flying in Levy is introduced for the current position of the seagull, P s (i) is the current position of the seagull, P bs (i) indicates the current optimal seagull position; Where r1 and r2 are random numbers between [0, 1] that follow a standard normal distribution, Levy (λ) is the random step length, λ is the power, and β = 1.5 here; In the later stage of iteration, as the seagulls tend to the optimal position as a whole, if the individuals in the optimal position are located near the local extreme value of the problem, the solution will fall into the local extreme value. Therefore, the t-distribution variation is introduced to perturb the optimal individual position of the seagulls, which helps to jump out of the local extreme value and improve the global optimization ability of the seagull optimization algorithm. in, represents the individual position of the seagull after the position of the current optimal seagull is disturbed by the t distribution, t(itre) represents the t distribution based on the number of iterations itre as the degree of freedom, P bs (i) indicates the current optimal seagull position; When f t <f x When , it means that the new position obtained after the t-distribution mutation disturbance is better than the fitness of the current best seagull. The individual with better fitness is retained as the best seagull and added to the seagull group, and the seagull with the worst fitness is eliminated. t Represents the fitness value of the new position, f x Represents the original fitness value; Step 3: Get multiple prediction parameters, and send the prediction parameters to the fuzzy fusion algorithm to perform data fusion on the prediction parameters; Step 4: Obtain the fusion result of the prediction data and predict the operation status of the hazardous chemicals transportation vehicle at the next moment; Step 5: Send the predicted results to the vehicle terminal for display.

2. A method for predicting and fusing data information of hazardous chemicals transportation vehicles according to claim 1, characterized in that: The various sensors described in step 1 include an acceleration sensor, a vehicle speed sensor, and a posture sensor installed on the vehicle body, and a gas concentration sensor and a temperature and humidity sensor installed on or inside the tank body. The acceleration sensor, vehicle speed sensor, posture sensor, gas concentration sensor and temperature and humidity sensor of the tank body are connected to the sensor group module signal of the microprocessor through a signal line or a wireless module; the microprocessor is also connected to a positioning module and an alarm module.

3. The method for predicting and fusing data information of hazardous chemicals transportation vehicles according to claim 1 is characterized by: The specific operation steps of the improved Seagull algorithm are: Step 1: Initialize algorithm parameters, including number of iterations, population size, problem dimension, and optimization range; Step 2: Initialize the individual positions of seagulls and calculate the fitness value of each seagull; step3: start iteration; Step 4: Calculate the optimal individual fitness of the seagull population and record the position of the optimal individual; Step 5: Generate candidate solutions based on the gray wolf hierarchy around the seagulls with better fitness, and calculate the fitness value of the corresponding solution; Step 6: Generate random numbers, perform spiral flight and Levy flight migration behaviors of seagulls based on the random number value and the probability of 0.5, and calculate the fitness value of each individual seagull; Step 7: Update the location information and fitness value of the optimal seagull; Step 8: Perturb the position of the optimal seagull according to the t distribution formula, compare the fitness of the mutated individual with that of the optimal seagull, retain the individual with better fitness as the optimal seagull, add it to the seagull group, and eliminate the seagull with the worst fitness; Step 9: When the number of iterations does not exceed the maximum number of iterations, repeat step 4 to step 8, otherwise the iteration loop ends; Step 10: Output the optimal seagull position and fitness value.

4. A method for predicting and fusing data information of hazardous chemicals transport vehicles according to claim 3, characterized in that: The weights and thresholds of the neural network are optimized by the improved Seagull algorithm, and the BP neural network is optimized by LGSOA to improve the prediction performance of the neural network. The main steps are: Step a: Initialize the parameters, the number of seagulls, and encode the weights and thresholds of the BP neural network to be optimized as the initial seagull population; Step b: Initialize the position of the seagull and process its position using a formula to prevent position overlap; Step c: Calculate the fitness of all current seagulls and find the best one as the best seagull in this iteration. The single output fitness is calculated using the root mean square error, and the multi-output fitness is the Euclidean distance error. Step d: Update the position of each seagull according to the formula; Step e: Determine whether the number of iterations has been reached, if yes, output the final position as the optimal seagull position, otherwise return to step c; Step f: Decode the optimal output into the initial weights and thresholds of the BP neural network, and train the neural network until the requirements are met.

5. A method for predicting and fusing data information of hazardous chemicals transport vehicles according to claim 4, characterized in that: After the LGSOA is used to optimize the BP neural network, a prediction alarm is performed for the single sensor data, and the obtained data is sent to the fuzzy fusion algorithm to fuse the prediction results, thereby predicting the operating status of the hazardous chemicals transport vehicle at the next moment.

Images