Intelligent diagnosis and early warning method for potential safety hazards of construction quality of ship final assembly

Abstract

The invention relates to the technical field of ship construction, and discloses an intelligent diagnosis and early warning method for ship final assembly construction quality potential safety hazards. The method comprises the following steps: receiving real-time sensing data of a plurality of detection nodes in ship construction, and analyzing the real-time sensing data through a pre-trained quality feature extraction model to generate a hidden danger feature vector and a risk level; and dynamically generating a hidden danger diagnosis strategy according to the sequence, wherein the hidden danger diagnosis strategy comprises a detection equipment scheduling instruction and a defect repair priority sequence. And utilizing the pre-trained hidden danger ship dynamic model to predict a defect diffusion trend, adjusting a dynamic response threshold value, optimizing, updating and repairing a priority sequence through an adaptive genetic algorithm, and outputting a quality early warning instruction to a ship construction management system. The method can accurately diagnose and early warn potential quality safety hazards, reasonably dispatch resources, improve the ship construction quality and safety, reduce the cost, and promote the intelligent development of the ship construction industry.

Description

Technical Field

[0001] The present invention relates to the technical field of shipbuilding, and particularly to an intelligent diagnosis and early warning method for potential quality and safety hazards in ship general assembly construction. Background Art

[0002] Ship general assembly construction is a complex and highly systematic project, involving numerous links and a huge technical system. During the shipbuilding process, the effective diagnosis and early warning of potential quality and safety hazards are of crucial importance, which is directly related to the navigation safety, service life, and economic benefits of the ship. However, the traditional methods for shipbuilding quality and safety detection and early warning have many limitations.

[0003] In the early stage, the quality inspection of shipbuilding mainly relied on manual experience. Inspectors judged whether there were defects in the ship structure, whether the welding was qualified, etc. by means of visual inspection and simple tool measurement. This method was not only inefficient but also greatly affected by human factors. There were differences in the technical levels and working states of different inspectors, and it was difficult to unify the judgment criteria for problems, easily missing some potential quality and safety hazards. For example, for the tiny cracks in the internal structure of the ship or the internal defects at the welding joints, it was very difficult for manual inspection to accurately detect them, and these hidden dangers might gradually expand during the long-term operation of the ship, leading to serious safety accidents.

[0004] With the development of technology, some simple instrument devices have been applied to shipbuilding inspection, such as ultrasonic flaw detectors, magnetic particle flaw detectors, etc. These devices have improved the accuracy and efficiency of inspection to a certain extent and can detect some defects that are difficult to find manually. However, they still have obvious deficiencies. They can only detect specific types of defects and cannot comprehensively analyze various parameters during the shipbuilding process. During the shipbuilding process, structural stress, welding deformation, material defects, and environmental factors are interrelated and interact with each other. A single inspection device cannot comprehensively evaluate the overall impact of these factors on the ship's quality and safety.

[0005] In addition, the traditional early warning methods often rely on fixed threshold settings. Once the detected data exceeds the preset threshold, a warning signal is issued. However, in actual shipbuilding, under different working conditions and environmental conditions, the development trends and hazard degrees of potential hazards are different. Fixed thresholds cannot be dynamically adjusted according to real-time situations, easily resulting in false alarms or missed alarms. For example, in a harsh weather environment, the stress borne by the ship structure will change, and the originally appropriate stress threshold may no longer be applicable. At this time, the fixed threshold early warning method is difficult to accurately judge potential quality and safety hazards.

[0006] In the trend of the increasing scale and complexity of shipbuilding projects, traditional methods for diagnosing and warning of quality and safety hazards can no longer meet the needs of modern shipbuilding. On the one hand, the construction cycle of large ships is long and the cost is high, and any quality and safety problems may lead to huge economic losses; on the other hand, with the development of the ship transportation industry, the requirements for the safety and reliability of ships are getting higher and higher. Therefore, there is an urgent need for an intelligent method that can comprehensively analyze various data, dynamically evaluate the risk of potential hazards in real time, and can accurately warn and efficiently handle quality and safety hazards to ensure the quality and safety of ship assembly construction. Summary of the Invention

[0007] The purpose of the present invention is to provide an intelligent diagnosis and warning method for quality and safety hazards in ship assembly construction to solve the problems raised in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solution: an intelligent diagnosis and warning method for quality and safety hazards in ship assembly construction, the method comprising: Receiving real-time sensing data of multiple detection nodes during the ship construction process, the sensing data including structural stress, welding deformation, material defect parameters, and environmental monitoring indicators; Based on a pre-trained quality feature extraction model, performing multi-dimensional fusion analysis on the real-time sensing data to generate potential hazard feature vectors and risk levels for each detection node; According to the risk level, dynamically generating a potential hazard diagnosis strategy, the strategy including detection equipment scheduling instructions and a defect repair priority sequence; Based on a pre-trained potential hazard propagation dynamics model, predicting the diffusion trend of defects within a preset future time period and adjusting the dynamic response threshold in the potential hazard diagnosis strategy; Optimizing the dynamic response threshold through an adaptive genetic algorithm, updating the defect repair priority sequence, and outputting the final quality warning instruction to the ship construction management system.

[0009] Preferably, the construction steps of the quality feature extraction model include: Collecting historical ship construction defect data to construct a training data set including defect types, geometric dimensions, and environmental correlation factors; Performing spatial feature extraction on the training data set through a residual neural network to generate an initial defect feature map; Combining an attention mechanism to perform weight assignment on the initial defect feature map to generate a multi-scale fusion feature; Combining the multi-scale fusion feature with a time series analysis module to generate the quality feature extraction model that can be updated online.

[0010] Preferably, the dynamically generating a potential hazard diagnosis strategy includes: Allocate initial repair weights to each detection node according to the risk level; Calculate the repair efficiency coefficient of each node based on real-time device availability data and personnel scheduling status; Non-linearly superimpose the repair efficiency coefficient and the initial repair weight to generate the defect repair priority sequence; Generate the detection device scheduling instruction according to the priority sequence result, in combination with the device load balancing constraint.

[0011] Preferably, the construction steps of the hidden danger propagation dynamics model include: Collect historical defect diffusion data and environmental disturbance data of ship section construction to construct a spatio-temporal evolution data set; Perform frequency domain feature decomposition on the spatio-temporal evolution data set through the wavelet transform algorithm to extract the critical frequency component of defect diffusion; Combine finite element simulation data to establish a structural stress propagation matrix to quantify the defect correlation strength between different detection nodes; Input the critical frequency component and the defect correlation strength into the temporal convolutional network to generate the hidden danger propagation dynamics model.

[0012] Preferably, the method further includes: Identify detection nodes whose diffusion rate exceeds a preset threshold according to the prediction result of the hidden danger propagation dynamics model; Assign high-risk marks to the identified nodes in the hidden danger diagnosis strategy; Based on the high-risk marks, trigger a redundant detection resource allocation mechanism and suspend low-priority detection tasks.

[0013] Preferably, the calculation of the repair efficiency coefficient includes: Obtain the device failure rate, personnel skill level, and environmental disturbance factor to construct a dynamic performance evaluation matrix; Calculate the optimal resource scheduling path between each node through the improved Dijkstra algorithm to generate the basic repair performance value; Perform a tensor product operation on the dynamic performance evaluation matrix and the basic repair performance value to obtain the repair efficiency coefficient; Among them, the calculation formula of the repair efficiency coefficient is:

[0014] In the formula, Q represents the repair efficiency coefficient, ρ j represents the reciprocal of the real-time failure rate of the j-th type of device, σ j represents the skill level weight of the j-th type of personnel, ϵ b represents the basic repair performance value, and m represents the total number of device categories.

[0015] Preferably, the generation of the multi-scale fusion features includes: Performing channel dimension normalization processing on the initial defect feature map to eliminate the dimensional difference; Extracting defect saliency features at different scales through a multi-head attention mechanism; Using a feature pyramid network to perform cross-scale fusion on the saliency features and reconstructing a hierarchical defect representation.

[0016] Preferably, the execution of the adaptive genetic algorithm includes: Setting an objective function for optimizing the defect repair priority, the function including repair timeliness and resource consumption constraints; Performing fitness evaluation on population individuals through an elitist retention strategy and dynamically adjusting the crossover and mutation probabilities; In each iteration, screening the non-dominated solution set based on the Pareto front and updating the dynamic response threshold; Outputting the quality warning instruction that meets the multi-objective convergence condition; Wherein, the objective function is:

[0017] In the formula, G represents the objective function value, R t represents the repair timeliness score, C r represents the resource consumption score, and λ and μ are respectively the normalization coefficients of timeliness and resource consumption.

[0018] Preferably, the method further includes: After allocating the high-risk mark, monitoring the defect repair progress of the corresponding node in real time; If the repair progress lags behind the preset threshold, activating the dynamic resource reallocation mechanism, increasing the number of detection devices and adjusting the personnel configuration.

[0019] Preferably, the setting of the resource consumption constraint includes: Collecting the energy consumption data of each detection device and the personnel man-hour cost, and constructing a resource consumption baseline model; Calculating the upper limit of the dynamic resource quota according to the real-time task load rate; Taking the upper limit of the dynamic resource quota as the constraint condition of the objective function; Wherein, the calculation formula of the upper limit of the dynamic resource quota is:

[0020] In the formula, C max represents the upper limit of the dynamic resource quota, η represents the energy consumption coefficient of the device per unit time, P dIt represents the real-time task load rate, τ is the labor cost conversion factor, and ω is the emergency resource reserve constant.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows: The intelligent diagnosis and early warning method for quality and safety hazards in ship general assembly construction proposed by the present invention has significant beneficial effects in many aspects. In terms of data processing and hazard diagnosis, the method receives real-time sensing data from multiple detection nodes during ship construction, including structural stress, welding deformation, material defect parameters, and environmental monitoring indicators, and uses a pre-trained quality feature extraction model for multi-dimensional fusion analysis. This process can fully explore the potential connections between various types of data. Compared with traditional methods that only rely on single or a small amount of data for judgment, it greatly improves the extraction accuracy of quality and safety hazard features. For example, traditional methods may only focus on a single piece of data on welding deformation and ignore the influence of structural stress and environmental factors on it, while the present invention can comprehensively consider these factors and more accurately judge whether there are quality and safety hazards at the welding part. The generated hazard feature vectors and risk levels are also more accurate and reliable, providing a solid data basis for subsequent processing.

[0022] In terms of diagnosis strategy generation and resource scheduling, dynamically generating hazard diagnosis strategies according to risk levels has great advantages. Initial repair weights are assigned to each detection node, and the repair efficiency coefficient is calculated by combining real-time device availability data and personnel scheduling status, and then a defect repair priority sequence is generated. This method fully considers the actual repair resources and personnel situation. Different from traditional methods of unified repair order or random repair arrangement, the present invention can give priority to processing nodes with high risk and high repair efficiency, effectively improving the overall efficiency of the repair work. At the same time, detection device scheduling instructions are generated in combination with the device load balancing constraint, avoiding overuse or idleness of the devices, improving the utilization rate of the detection devices, reducing the device maintenance cost, and making the resources more reasonably configured.

[0023] In terms of hazard prediction and dynamic adjustment, predicting the diffusion trend of defects within a preset future time period based on a pre-trained hazard propagation dynamics model and adjusting the dynamic response threshold has strong foresight. By predicting the defect diffusion in advance, corresponding measures can be taken when the hazards have not yet expanded, such as strengthening detection, adjusting the construction process, etc., effectively avoiding serious quality and safety problems caused by defect diffusion. And the dynamic response threshold is optimized using an adaptive genetic algorithm, further improving the accuracy and timeliness of early warning. During the repair process, the dynamic resource reallocation mechanism is monitored and activated in real time according to the repair progress, ensuring the smooth progress of the repair work, avoiding repair delays caused by insufficient resources, and guaranteeing the quality and progress of ship construction.

[0024] Overall, this method improves the quality and safety of shipbuilding, reduces the risks of ship failures and accidents caused by quality and safety issues, and extends the service life of ships. Economically, it effectively reduces the costs of shipbuilding and subsequent maintenance, avoids the additional costs brought by rework due to quality problems, and reduces the operating losses caused by ship failures. In terms of industry development, it promotes the development of shipbuilding technology towards intelligence and high efficiency, providing strong technical support for the sustainable development of the shipbuilding industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 FIG. is the working principle diagram of the intelligent diagnosis and early warning method for quality and safety hazards in ship assembly construction according to the present invention; Figure 2 FIG. is the construction step diagram of the hazard propagation dynamics model; Figure 3 FIG. is the flow chart of the treatment measures based on the prediction results; Figure 4 FIG. is the monitoring flow chart of the repair progress of high-risk nodes. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] Please refer to Figures 1-4 , the present invention provides an intelligent diagnosis and early warning method for quality and safety hazards in ship assembly construction, aiming to achieve efficient diagnosis and accurate early warning of quality and safety hazards in the shipbuilding process. The specific steps are as follows: Receiving real-time sensing data: During the shipbuilding process, sensors are arranged at multiple key detection nodes to collect various real-time sensing data. These detection nodes are distributed in different structural parts of the ship, covering areas that have an important impact on the quality and safety of the ship. The collected sensing data includes structural stress data for monitoring the magnitude and distribution of stress borne by the ship structure during construction; welding deformation data that can reflect the deformation caused by the welding process to the structure; material defect parameters for detecting whether there are defects inside the material and their characteristics; and environmental monitoring indicators, such as environmental factor data such as temperature, humidity, and wind speed at the construction site. Through the data transmission network, these real-time sensing data are transmitted to the data processing system in real time.

[0028] Hidden danger feature vector and risk level generation: Use a pre-trained quality feature extraction model to process the received real-time sensing data. This model is trained based on historical shipbuilding data and has strong feature extraction capabilities. It performs multi-dimensional fusion analysis on the real-time sensing data, comprehensively considering the correlations and features between different types of data, and generates hidden danger feature vectors for each detection node. At the same time, according to the features of the hidden danger feature vectors and the pre-set risk assessment rules, determine the risk level of each detection node. The risk level can be divided into different levels such as low risk, medium risk, and high risk, so as to take targeted measures subsequently.

[0029] Dynamic generation of hidden danger diagnosis strategy: Dynamically generate a hidden danger diagnosis strategy according to the determined risk level. This strategy includes two key parts: detection device scheduling instructions and defect repair priority sequences. For the detection device scheduling instructions, according to the risk levels of each detection node and the current distribution and availability status of the detection devices, reasonably arrange the detection devices to go to the corresponding nodes for further detection to ensure the efficient use of resources. For the defect repair priority sequences, comprehensively consider factors such as risk level, repair difficulty, and impact on the overall quality of the ship, and determine the repair order for each detection node with defects to ensure that the defects with greater impact on the ship's quality and safety are processed first.

[0030] Prediction of defect diffusion trend and strategy adjustment: With the help of a pre-trained hidden danger propagation dynamics model, predict the defect diffusion trend within a preset future time period. This model combines historical defect diffusion data and the actual situation of current shipbuilding to simulate and predict the diffusion of defects under different conditions. According to the prediction results, adjust the dynamic response threshold in the hidden danger diagnosis strategy. If it is predicted that the defect has a trend of rapid diffusion, then lower the dynamic response threshold to discover and handle potential problems more timely; conversely, if it is predicted that the defect diffuses slowly, then appropriately increase the dynamic response threshold to reasonably allocate resources.

[0031] Optimization of dynamic response threshold and output of early warning instructions: Optimize the dynamic response threshold through an adaptive genetic algorithm. The adaptive genetic algorithm takes repair timeliness and resource consumption as constraint conditions and continuously adjusts the dynamic response threshold to achieve the best diagnosis and early warning effects. During the optimization process, update the defect repair priority sequence to ensure its more reasonableness. Finally, generate quality early warning instructions based on the optimized results and output them to the shipbuilding management system. After receiving the early warning instructions, the shipbuilding management system can take corresponding measures in a timely manner, such as arranging maintenance personnel for repair and adjusting the construction progress, to ensure the quality and safety of shipbuilding.

[0032] The following further illustrates the implementation of the present invention in combination with Embodiments 1 to 6.

[0033] Embodiment 1:

[0034] In this embodiment, the construction of the quality feature extraction model and the generation process of multi-scale fusion features are mainly elaborated.

[0035] The construction of the quality feature extraction model is crucial for accurately diagnosing potential quality and safety hazards in shipbuilding. First, historical shipbuilding defect data are widely collected. These data cover various types of defects, including but not limited to welding defects, internal cracks in materials, etc. At the same time, the geometric dimensions of the defects are recorded, such as length, width, depth, etc., as well as the associated factors related to the environment, such as environmental temperature, humidity during construction, current and voltage during welding, etc. A training data set containing defect types, geometric dimensions, and environmental associated factors is constructed through these data.

[0036] Next, a residual neural network is used to extract spatial features from the training data set. The residual neural network has strong feature learning ability and can effectively extract spatial features in the data. In this process, it performs layer-by-layer convolution operations on the input training data, gradually extracting deep spatial features and generating an initial defect feature map. This initial defect feature map contains the basic feature information of the defect in the spatial dimension, but it is not comprehensive and accurate enough.

[0037] To further optimize the features, the attention mechanism is combined to assign weights to the initial defect feature map. The attention mechanism can automatically focus on the important feature parts in the data and assign higher weights to these key features. When dealing with shipbuilding defect data, it can highlight the features that are more critical for defect diagnosis, such as the edge features of the defect, abnormal features at specific positions, etc., thus generating multi-scale fusion features.

[0038] When generating multi-scale fusion features, the specific operations are as follows: First, the initial defect feature map is normalized in the channel dimension. Since the data in different channels may have different dimensions and numerical ranges, the normalization process can eliminate the dimension differences, making different features comparable. Then, the defect saliency features at different scales are extracted through the multi-head attention mechanism. The multi-head attention mechanism analyzes the data from multiple different perspectives and can extract richer feature information. Finally, the feature pyramid network is used to perform cross-scale fusion on these saliency features to reconstruct the hierarchical defect representation. The feature pyramid network can effectively fuse the features at different scales to form a more comprehensive and representative defect feature representation, providing more accurate data support for subsequent hazard diagnosis.

[0039] Combine the multi-scale fusion features with the time series analysis module to generate a quality feature extraction model that can be updated online. The time series analysis module can capture the changing trend of data over time. Considering that the shipbuilding process is a dynamic process, the data changes at different time points are also of great significance for potential hazard diagnosis. By combining the time series analysis module with the multi-scale fusion features, the quality feature extraction model can continuously adapt to new data, achieve online update, and improve the accuracy and timeliness of potential hazard diagnosis for shipbuilding quality and safety.

[0040] Example 2:

[0041] When dynamically generating a potential hazard diagnosis strategy, first assign an initial repair weight to each detection node according to its risk level. Nodes with a high risk level are assigned a higher initial repair weight, and nodes with a low risk level are assigned a lower initial repair weight, which can ensure that defects with greater risks are processed first.

[0042] Then, based on the real-time device availability data and personnel scheduling status, calculate the repair efficiency coefficient of each node. Obtain the equipment failure rate, personnel skill level, and environmental interference factor, and construct a dynamic efficiency evaluation matrix. The equipment failure rate reflects the possibility of the equipment malfunctioning under the current usage state. The lower the equipment failure rate, the higher the reliability of its normal operation during the repair process, and the greater the positive impact on the repair efficiency. The weight of the personnel skill level reflects the contribution degree of personnel with different skill levels to the repair work. The higher the skill level of the personnel, the higher the possible repair efficiency and quality. The environmental interference factor takes into account the impact of the construction site environment on the repair work. For example, bad weather conditions may reduce the repair efficiency.

[0043] Calculate the optimal resource scheduling path between each node through the improved Dijkstra algorithm to generate the basic repair efficiency value. The Dijkstra algorithm is a classic shortest path algorithm, which is used in this scenario to find the optimal path from the resource supply point to each detection node to determine the best way of resource scheduling, so as to obtain the basic repair efficiency value. The basic repair efficiency value reflects the repair efficiency that can be achieved by resource scheduling under ideal conditions without considering the equipment failure rate, personnel skill level, and environmental interference factor.

[0044] Perform a tensor product operation on the dynamic efficiency evaluation matrix and the basic repair efficiency value to obtain the repair efficiency coefficient. The calculation formula for the repair efficiency coefficient is:

[0045] where Q represents the repair efficiency coefficient, which comprehensively reflects the influence of various factors such as equipment, personnel, and environment on the repair efficiency. ρ jDenotes the reciprocal of the real-time failure rate of the j-th type of equipment. The larger the reciprocal of the real-time failure rate, the more reliable the equipment, and the greater the improvement effect on the repair efficiency. σ j Denotes the skill level weight of the j-th type of personnel, which is used to measure the contribution of personnel with different skill levels to the repair efficiency. ϵ b Denotes the basic repair efficiency value, which is calculated by the improved Dijkstra algorithm. m represents the total number of equipment categories, covering various types of equipment used in the shipbuilding process.

[0046] Non-linearly superimpose the repair efficiency coefficient and the initial repair weight to generate the defect repair priority sequence. This non-linear superposition takes into account two aspects of factors, namely the repair weight and the repair efficiency, making the repair priority sequence more reasonable. For nodes with high repair efficiency and large initial repair weights, their positions in the repair priority sequence are more forward.

[0047] Finally, according to the results of the priority sequence, combined with the equipment load balancing constraint, generate the detection equipment scheduling instruction. When generating the scheduling instruction, it is necessary to ensure that high-priority nodes are detected first, and at the same time consider the load balancing of the equipment, avoiding overuse of some equipment while other equipment is idle, and improving the overall utilization efficiency of the detection equipment.

[0048] Example 3:

[0049] This example elaborates in detail the construction process of the hidden danger propagation dynamics model. The hidden danger propagation dynamics model is of great significance for predicting the diffusion trend of defects and taking preventive measures in advance.

[0050] Widely collect historical defect diffusion data and environmental perturbation data of ship section construction. The historical defect diffusion data records the diffusion of defects over time and environmental changes during past shipbuilding processes, including information such as the diffusion speed, direction, and range of defects. The environmental perturbation data contains data on changes in environmental factors during the construction process, such as temperature fluctuations, humidity changes, and external forces, which affect the diffusion of defects. Construct a spatio-temporal evolution data set through these data.

[0051] Perform frequency-domain feature decomposition on the spatio-temporal evolution data set through the wavelet transform algorithm. The wavelet transform algorithm can decompose complex time series data into components of different frequencies and extract the critical frequency components of defect diffusion. These critical frequency components reflect the key frequency characteristics during the defect diffusion process and play an important role in understanding the internal mechanism and law of defect diffusion.

[0052] Combined with finite element simulation data, a structural stress propagation matrix is established. Finite element simulation discretizes the ship structure to simulate the stress distribution and changes of the structure under various working conditions. According to the simulation results, the stress transfer relationship between different detection nodes is calculated to quantify the defect correlation strength between different detection nodes. The defect correlation strength represents the degree to which the defect of one node affects other nodes. The greater the correlation strength, the higher the possibility of defect propagation between the two nodes.

[0053] The critical frequency component and the defect correlation strength are input into the temporal graph convolutional network. The temporal graph convolutional network combines the advantages of graph convolutional network and time series analysis and can effectively process data with spatio-temporal characteristics. In this embodiment, it uses the critical frequency component and the defect correlation strength to learn the propagation law of defects in the time and space dimensions and generates a hidden danger propagation dynamics model. This model can accurately predict the diffusion trend of defects within a preset future time period, providing a strong basis for adjusting the hidden danger diagnosis strategy.

[0054] Embodiment 4:

[0055] This embodiment mainly introduces the relevant processing measures based on the prediction results of the hidden danger propagation dynamics model, as well as the monitoring of the repair progress of high-risk nodes and the dynamic resource reallocation mechanism. Specifically, it includes: According to the prediction results of the hidden danger propagation dynamics model, the detection nodes with a diffusion rate exceeding the preset threshold are identified. The preset threshold is set based on a large amount of historical data and experience and is used to judge the severity of defect diffusion. When it is detected that the defect diffusion rate of a certain node exceeds the preset threshold, it indicates that the defect of this node has a greater potential risk and needs to be focused on.

[0056] High-risk marks are assigned to the identified nodes in the hidden danger diagnosis strategy. High-risk marks can significantly distinguish these high-risk nodes so that special measures can be taken in subsequent processing. For example, when formulating the detection plan and repair plan, these high-risk nodes are given priority.

[0057] Based on the high-risk marks, a redundant detection resource allocation mechanism is triggered, and low-priority detection tasks are suspended. The redundant detection resource allocation mechanism can increase the detection frequency and means for high-risk nodes to ensure timely detection of changes in defects. At the same time, suspending low-priority detection tasks can concentrate resources to handle the problems of high-risk nodes and improve resource utilization efficiency.

[0058] After assigning high-risk tags, the defect repair progress of the corresponding nodes is monitored in real time. The repair progress data is obtained by installing monitoring equipment at the repair site or requiring maintenance personnel to report the repair situation regularly. The actual repair progress is compared with the preset threshold. If the repair progress lags behind the preset threshold, it indicates that there are problems with the repair work, and there may be situations such as greater repair difficulty and insufficient resources.

[0059] At this time, activate the dynamic resource reallocation mechanism, increase the number of detection devices and adjust the personnel configuration. Increasing the detection devices can improve the accuracy and efficiency of detection and timely discover new problems that may occur during the repair process. Adjusting the personnel configuration can, according to the actual situation, deploy more experienced or more skill-matched personnel to participate in the repair work, speed up the repair progress, and ensure the quality and safety of shipbuilding.

[0060] Example 5:

[0061] During the execution of the adaptive genetic algorithm, first set the objective function for optimizing the defect repair priority. The objective function is:

[0062] Among them, G represents the value of the objective function, which comprehensively considers two key factors: repair timeliness and resource consumption. R t represents the repair timeliness score, which is used to measure the efficiency of the repair work in terms of time. The higher the repair timeliness score, the faster the repair work is completed. C r represents the resource consumption score, which reflects the resources consumed during the repair process, such as the energy consumption of detection devices and the man-hour cost of maintenance personnel. The lower the resource consumption score, the higher the resource utilization efficiency. λ and μ are the normalization coefficients of timeliness and resource consumption respectively, which are used to adjust the relative importance of repair timeliness and resource consumption in the objective function and can be reasonably set according to the actual situation.

[0063] Through the elitist retention strategy, the fitness of population individuals is evaluated, and the crossover and mutation probabilities are dynamically adjusted. The elitist retention strategy ensures that in the iterative process of the genetic algorithm, individuals with higher fitness in each generation can be directly retained in the next generation, avoiding the loss of excellent individuals. When evaluating the fitness of population individuals, the fitness value of each individual is calculated according to the objective function. The higher the fitness value of an individual, the greater the probability of being selected in the next generation. At the same time, according to the fitness distribution of population individuals, the crossover and mutation probabilities are dynamically adjusted. If the fitness differences of population individuals are small, it indicates that the algorithm may fall into a local optimal solution. At this time, appropriately increase the mutation probability to increase the diversity of the population and help the algorithm jump out of the local optimum; if the fitness differences of population individuals are large, then appropriately reduce the mutation probability to speed up the convergence rate of the algorithm.

[0064] In each iteration, the non-dominated solution set is screened based on the Pareto front. The Pareto front is a set of solutions in a multi-objective optimization problem where no solution can improve at least one objective without sacrificing other objectives. By screening the non-dominated solution set, a series of solutions that achieve a better balance between repair timeliness and resource consumption can be obtained. Select appropriate solutions from these non-dominated solution sets to update the dynamic response threshold.

[0065] Continue the iteration until the multi-objective convergence condition is met. The multi-objective convergence condition can be set according to actual needs. For example, when the objective function values change very little within a certain number of iterations, or when the non-dominated solution set no longer changes significantly, it is considered that the algorithm has converged. When the convergence condition is satisfied, output a quality warning instruction that meets the multi-objective convergence condition to provide an accurate decision-making basis for the shipbuilding management system.

[0066] Example 6:

[0067] This example mainly introduces the setting process of resource consumption constraints and the calculation method and application of the dynamic resource quota upper limit.

[0068] When setting resource consumption constraints, first collect the energy consumption data of each detection device and the labor cost of personnel. The energy consumption data of the detection device is obtained by installing an energy consumption monitoring device on the device, and the energy consumption of the device in different working states is recorded. The labor cost of personnel is calculated based on factors such as the salary standard and working hours of maintenance personnel. Through these data, a resource consumption baseline model is constructed. The resource consumption baseline model reflects the resource consumption level required to complete certain detection and repair tasks under normal circumstances.

[0069] Then, calculate the dynamic resource quota upper limit according to the real-time task load rate. The real-time task load rate reflects the busy degree of the current shipbuilding task. The higher the task load rate, the more resources need to be invested. The calculation formula for the dynamic resource quota upper limit is:

[0070] where, C max represents the dynamic resource quota upper limit, which limits the maximum amount of resources allowed to be consumed under the current task load rate. η represents the energy consumption coefficient per unit time of the device, which is used to measure the difference in energy consumption levels of different devices. P d represents the real-time task load rate, which is calculated based on the actual situation of the current shipbuilding task. τ is the labor cost conversion factor, which is used to convert the labor cost of personnel into a resource measurement unit unified with the energy consumption of the device. ω is the emergency resource reserve constant, which reserves a certain amount of resources to cope with emergencies.

[0071] Take the upper limit of the dynamic resource quota as a constraint condition for the objective function. During the optimization process of the adaptive genetic algorithm, ensure that the generated repair plan does not exceed the upper limit of the dynamic resource quota in terms of resource consumption. This can effectively control the use of resources, avoid excessive resource consumption, and ensure the economy and sustainability of the shipbuilding process. At the same time, when formulating the detection equipment scheduling instructions and defect repair plans, also based on the upper limit of the dynamic resource quota, reasonably arrange resources, and improve the resource utilization efficiency.

[0072] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0073] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent diagnosis and early warning method for potential quality and safety hazards in ship general assembly construction, characterized in that, It includes: Receiving real-time sensing data of multiple detection nodes during shipbuilding, where the sensing data includes structural stress, welding deformation, material defect parameters, and environmental monitoring indicators; Based on a pre-trained quality feature extraction model, performing multi-dimensional fusion analysis on the real-time sensing data to generate hidden danger feature vectors and risk levels for each detection node; According to the risk level, dynamically generating a hidden danger diagnosis strategy, where the strategy includes detection equipment scheduling instructions and a defect repair priority sequence; Based on a pre-trained hidden danger propagation dynamics model, predicting the diffusion trend of defects within a preset future time period and adjusting the dynamic response threshold in the hidden danger diagnosis strategy; Optimizing the dynamic response threshold through an adaptive genetic algorithm, updating the defect repair priority sequence, and outputting the final quality warning instruction to the shipbuilding management system.

2. The intelligent diagnosis and early warning method according to claim 1, wherein The construction steps of the quality feature extraction model include: Collecting historical shipbuilding defect data and constructing a training data set including defect types, geometric dimensions, and environmental correlation factors; Performing spatial feature extraction on the training data set through a residual neural network to generate an initial defect feature map; Combining an attention mechanism to assign weights to the initial defect feature map to generate multi-scale fusion features; Combining the multi-scale fusion features with a time series analysis module to generate the quality feature extraction model that can be updated online.

3. The intelligent diagnosis and early warning method according to claim 1, wherein The dynamically generating hidden danger diagnosis strategy includes: According to the risk level, assigning an initial repair weight to each detection node; Based on real-time device availability data and personnel scheduling status, calculating the repair efficiency coefficient of each node; Non-linearly superimposing the repair efficiency coefficient and the initial repair weight to generate the defect repair priority sequence; According to the priority sequence result, combining with the equipment load balancing constraint, generating the detection equipment scheduling instruction.

4. The intelligent diagnosis and warning method according to claim 1, wherein The construction steps of the hidden danger propagation dynamics model include: Collecting historical defect diffusion data and environmental perturbation data of ship section construction to construct a spatio-temporal evolution data set; Performing frequency domain feature decomposition on the spatio-temporal evolution data set through a wavelet transform algorithm to extract the critical frequency components of defect diffusion; Combining finite element simulation data to establish a structural stress propagation matrix to quantify the defect correlation strength between different detection nodes; Inputting the critical frequency components and the defect correlation strength into a temporal graph convolutional network to generate the hidden danger propagation dynamics model.

5. The intelligent diagnosis and early warning method according to claim 4, wherein It also includes: According to the prediction result of the hidden danger propagation dynamics model, identifying the detection nodes whose diffusion rate exceeds the preset threshold; Assigning a high-risk mark to the identified nodes in the hidden danger diagnosis strategy; Based on the high-risk mark, triggering a redundant detection resource allocation mechanism and suspending low-priority detection tasks.

6. The intelligent diagnosis and early warning method according to claim 3, characterized in that The calculation of the repair efficiency coefficient includes: Obtaining the equipment failure rate, personnel skill level, and environmental interference factor to construct a dynamic efficiency evaluation matrix; Calculating the optimal resource scheduling path between each node through an improved Dijkstra algorithm to generate a basic repair efficiency value; Performing a tensor product operation on the dynamic efficiency evaluation matrix and the basic repair efficiency value to obtain the repair efficiency coefficient; Among them, the calculation formula of the repair efficiency coefficient is as follows: ； Wherein, Q represents the repair efficiency coefficient, ρ j represents the reciprocal of the real-time failure rate of the j-th type of equipment, σ j represents the skill level weight of the j-th type of personnel, ϵ b represents the basic repair effectiveness value, and m represents the total number of equipment categories.

7. The intelligent diagnosis and early warning method according to claim 2, wherein The generation of the multi-scale fusion features includes: Performing channel dimension normalization processing on the initial defect feature map to eliminate the dimension difference; Extracting defect saliency features at different scales through a multi-head attention mechanism; Using a feature pyramid network to perform cross-scale fusion on the saliency features and reconstructing a hierarchical defect representation.

8. The intelligent diagnosis and warning method according to claim 1, wherein, The execution of the adaptive genetic algorithm includes: Setting an objective function for optimizing the defect repair priority, and the function includes repair timeliness and resource consumption constraints; Performing fitness evaluation on population individuals through an elitist retention strategy and dynamically adjusting the crossover and mutation probabilities; In each iteration, screening the non-dominated solution set based on the Pareto front and updating the dynamic response threshold; Outputting the quality warning instruction that meets the multi-objective convergence condition; Among them, the objective function is: ； where G represents the objective function value, R t represents the repair aging score, C r represents the resource consumption score, and λ and μ are the normalization coefficients of aging and resource consumption respectively.

9. The intelligent diagnosis and warning method according to claim 5, wherein, It also includes: After allocating the high-risk mark, the defect repair progress of the corresponding node is monitored in real time; If the repair progress lags behind the preset threshold, activate the dynamic resource reallocation mechanism, increase the number of detection devices and adjust the personnel configuration.

10. The intelligent diagnosis and early warning method according to claim 8, characterized in that, The setting of the resource consumption constraint includes: Collecting the energy consumption data of each detection device and the personnel working hour cost, and constructing a resource consumption baseline model; Calculating the upper limit of the dynamic resource quota according to the real-time task load rate; Taking the upper limit of the dynamic resource quota as the constraint condition of the objective function; Among them, the calculation formula of the upper limit of the dynamic resource quota is: ； Where C max represents the upper limit of the dynamic resource quota, η represents the energy consumption coefficient of the device per unit time, P d represents the real-time task load rate, τ is the conversion factor of the man-hour cost, and ω is the constant of the emergency resource reserve.

Images