336 results about "Path length" patented technology

The invention belongs to the technical field of path planning, particularly relates to a robot dog inspection path planning method and system, solves the problems of poor operation endurance and inspection efficiency of a robot dog in a path planning method in the prior art, and comprises the following steps: S1, obtaining a three-dimensional grid map of an inspection environment; s2, generating an inspection point sequence; s3, sequentially planning each path section based on an improved A * algorithm, wherein the actual cost # imgabs1 # from the starting point of the path section to the current node # imgabs0 # is a scalar value integrating the path length, the traffic energy consumption and the equipment inspection information value; the heuristic cost # imgabs3 # from the current node # imgabs2 # to the end point of the path section is in direct proportion to the Euclidean distance from the current node # imgabs4 # to the end point of the path section; s4, when an obstacle is detected, performing dynamic re-planning of a local path; and S5, setting an optimal marching gait parameter for each terrain road section. And the path length, the passing energy consumption and the information value of the inspection point are comprehensively considered, so that the operation endurance and the inspection efficiency of the robot dog are improved.

The invention discloses a precise mold design method and system based on an intelligent optimization algorithm, which can effectively balance a plurality of key performance indexes such as path length, cutting load distribution and surface quality by constructing a multi-objective optimization framework, combining iterative computation to generate a candidate path set and introducing a Pareto optimal solution screening mechanism. On this basis, path parameters are further dynamically adjusted in combination with a process experience database, and path inflection points and transition sections are optimized through a neighborhood disturbance strategy, so that the finally generated processing path not only meets the requirements of high efficiency and stability, but also can adapt to actual processing requirements under different process conditions; according to the method, the machining efficiency and the surface consistency of complex mold parts are remarkably improved, the trial and error cost is reduced, and intelligent and self-adaptive precision mold design and machining path planning are achieved.

The invention discloses a path planning and hierarchical cooperative control method and system for an unmanned aerial vehicle cluster. The system comprises a path planning module and a formation motion control module. The method comprises the steps that firstly, an improved RRT * algorithm is adopted by a path planning module, through a multi-strategy heuristic node expansion mechanism fusing target bias and artificial potential field guidance and comprehensively considering path length, channel volume and Z-axis height change, a center path and a three-dimensional safe channel which take into account safety and smoothness are planned for an unmanned aerial vehicle cluster; and then, based on the central path, the formation motion control module adopts a distributed model prediction control framework, designs different optimization targets for a navigator and a follower, and solves an optimal control instruction on line, so that a cluster is guided to complete trajectory tracking, collision avoidance among individuals and self-adaptive formation reconstruction in a secure channel. According to the invention, the navigation problem of the unmanned aerial vehicle cluster in a complex obstacle environment is solved, and the path planning efficiency and the robustness of cooperative control are improved.

A path planning method and apparatus, a computer device and a storage medium. The method comprises: acquiring an initial work position of a robot (S202); acquiring a current map, the current map comprising a plurality of candidate work regions (S204); determining within a passable region of the current map an initial transition path corresponding to each candidate work region, the initial transition path being used for indicating a path from the initial work position to the candidate work region (S206); on the basis of the region area of each candidate work region, and the path length and the deflection angle of the initial transition path corresponding to each candidate work region, determining a region arrival cost, so as to obtain the region arrival cost corresponding to each candidate work region (S208); on the basis of the region arrival costs, determining a target work region of the robot from the candidate work regions (S210); and continuing to determine a next target work region of the robot from a latest map until a first termination condition is satisfied, and terminating path planning (S212).

The invention provides an unmanned aerial vehicle pose visual angle optimization method and system for fracture refined shooting. The method comprises the steps of performing fracture detection and boundary extraction on a coarse inspection image; recovering a camera track and sparse point cloud based on multi-view three-dimensional reconstruction, carrying out back projection and estimating a normal vector of a crack surface; constructing a shooting spherical shell with limited inner and outer radiuses by taking the crack point as a center, and generating a view cone and spherical shell intersection region allowed to be shot as a candidate set in combination with a normal vector; sampling in the candidate area to generate a plurality of candidate shooting points, and synchronously resolving the flight and holder integrated pose of the corresponding unmanned aerial vehicle; constructing a multi-target cost function including path length, attitude, pan-tilt angle and shooting error, and generating an optimal shooting point sequence and an inspection path through an optimization algorithm; the system realizes fine, efficient and automatic shooting of cracks in a complex structure environment through cooperation of multiple modules, and effectively improves the imaging quality and the detection precision.

The invention relates to a power line network topology estimation method and device and computer equipment. The method comprises the following steps: sending a modulation signal, and receiving a plurality of reflection signals corresponding to the modulation signal; determining a path length set and a current topological structure according to the signal propagation time corresponding to each reflected signal; the signal propagation time is a time difference between the sending time of the modulation signal and the receiving time of the reflection signal; selecting an expansion node from the current topological structure, and updating the current topological structure according to the expansion node and the path length set; the extension nodes are unselected nodes in the current topological structure. According to the topology estimation method, the complexity and cost of system deployment are reduced through a single modem implementation method, the topology structure is updated by analyzing the expansion nodes, dynamic network changes are supported, the applicability of the topology estimation method is improved, and the efficiency of topology network management and maintenance is improved.

The invention discloses a safety-enhanced trajectory planning method based on an Astar algorithm, and belongs to the technical field of intelligent robot navigation and path planning, and the method comprises the following steps: 1, constructing a grid map, and marking an obstacle region; 2, establishing a security evaluation mechanism, evaluating the surrounding environment of the path node, and quantifying a security index; 3, in a node expansion process, combining a heuristic item and a security index in a cost function to realize priority search for a security region during path selection; 4, key nodes of the obtained initial safety path are extracted based on a Douglas-Peucker algorithm to generate a rectangular safety corridor, and smoothing processing of the initial path is achieved based on a segmented Bezier curve. On the basis of guaranteeing path optimality, safety is used as an auxiliary guiding basis in the path searching process, and the path searching efficiency is improved. Therefore, consideration of the path length and the environmental risk is realized, and the navigation path with higher execution safety is generated.

The invention provides an adaptive multi-view track roaming optimization method based on Cesium, and the method comprises the steps: obtaining an original path point, and carrying out the processing of the original path point, and generating a smooth path; dynamically adjusting the time rate of the smooth path movement; dynamically adjusting the number of interpolation points of the smooth path; loading the dynamically adjusted path, and binding with the path through a dual-mode view angle; and rendering the path in real time, and interacting with a user. According to the method, a Catmull-Rom spline interpolation and inflection point optimization method is adopted, so that turning is smoother; the terrain height is dynamically sampled, terrain adaptation is accurate, and the model is prevented from being suspended or falling into the ground; according to the invention, first person / third person switching is supported, and flexible visual angle control is realized; the speed calculation based on the path length ensures that the motion rhythm is natural.

The invention provides a teaching-free welding robot dynamic energy-saving path planning method and system, and relates to the technical field of industrial robos.The method comprises the steps that 1, a workpiece is scanned through a laser vision device, weld joint coordinates, obstacle information and workpiece thermal deformation data are obtained, weld joint feature points are extracted, and the weld joint feature points are obtained; establishing a mapping relation between a workpiece coordinate system and a robot base coordinate system; step 2, based on the mapping relationship, in combination with a thermal deformation prediction mechanism and sensor feedback data, implanting a dynamic response node in the initial path, calculating path point offset by using the feedback data, and generating an anti-offset path; and step 3, aiming at the anti-offset path, constructing a multi-dimensional collaborative evaluation mechanism, and carrying out synchronous adjustment and comprehensive balance on the path length, the energy consumption of the whole welding process and the welding quality parameters so as to obtain a multi-dimensional collaborative evaluation result. Weld joint deformation, clamp offset and obstacle displacement are sensed in real time, the path is dynamically planned, and the flexibility of path planning is improved.

The invention discloses an improved robot motion planning method based on Jump Point Search (JPS), and relates to a robot motion planning method based on JPS. According to the method, the search direction and the path planning process of the JPS algorithm are optimized, the path direction guidance of BFS (breadth-First Search) is combined, and the search direction priorities are sorted, so that the overall search efficiency and the path security are remarkably improved. Specifically, according to the method, a cosine value is used for measuring the similarity degree of each search direction and a BFS path, and the directions are divided into high priorities and secondary priorities. In the hop point searching process, the heuristic function value only evaluates the extended hop points in the high priority direction, and selects the hop point corresponding to the minimum value, so that the extension in the redundant direction is reduced, and the searching efficiency is improved. In addition, the invention also designs a set of path optimization strategy which mainly comprises two parts, namely a'security node updating strategy 'and a'redundant node eliminating strategy'. Aiming at the dangerous boundary path points, a security node updating strategy is utilized, and adjacent nodes are replaced by common security neighbor points, so that paths with potential safety hazards are prevented from being formed; according to the redundant node elimination strategy, the number of control nodes is reduced and the overall path length is shortened by detecting the direct connectivity between continuous nodes.

The invention discloses a path planning method based on a pheromone enhanced ant colony algorithm, and particularly relates to the technical field of path planning. By constructing an initial pheromone matrix, the ratio of the linear Euclidean distance between a path starting point and a path end point to the sum of the Euclidean distance from the path starting point to a candidate grid position and the Euclidean distance from the path end point to the candidate grid position is calculated, and the initial pheromone concentration is distributed near a theoretical optimal path; an elite solution pheromone enhancement strategy is adopted, all generated paths are sorted according to path quality in each iteration, and a high-quality path is selected, copied and then combined with a global optimal path to be used for updating a pheromone matrix; a prospective steering optimization strategy is utilized, punishment is applied to redundant turning in the searching process, and the pheromone updating amount is determined through the path length and the turning frequency; through multi-round iteration, stable pheromone distribution is realized, so that a final optimized path is obtained, the search efficiency is improved, and the method is suitable for a path planning task in a complex environment.

The invention relates to the technical field, in particular to an unmanned inspection method and system for a switch cabinet, and the method comprises the following steps: building an incidence relation between signal nodes through a graph neural network, improving the time sequence consistency of multi-source signals, enabling instrument reading and temperature and humidity signals to form a hierarchical structure under a unified time frame, and improving the accuracy of inspection. The logic association of state layering is enhanced, continuous sampling sequence features are extracted through a long-short-term memory network, the response sensitivity and time prediction capability to abnormal signal abrupt change are improved, amplitude matching of temperature difference and current signals is achieved under the same time index, early abnormal nodes are accurately recognized, and a dynamic positioning sequence is established. Through joint tradeoff of distance and energy consumption, path lengthens and energy peak values are reduced, task throughput and equipment availability are improved, timing sequence capture and trend judgment capabilities are enhanced in an anomaly recognition layer, and accuracy and continuity of inspection decision and reliability of safety assessment are integrally improved.

Machine learning techniques are used, in one embodiment, to mitigate multipath in an L5 GNSS receiver. In one embodiment, training data is generated to provide ground truth data for excess path length (EPL) corrections for a set of received GNSS signals. A system extracts features from the set of received GNSS signals and uses the extracted features and the ground truth data to train a set of one or more neural networks that can produce EPL corrections for pseudorange measurements. The trained set of one or more neural networks can be deployed in GNSS receivers and used in the GNSS receivers to correct pseudorange measurements using EPL corrections provided by the trained set of neural networks.

The invention relates to the technical field of key generation and distribution, in particular to a multi-device automatic configuration encryption method and system based on Bluetooth. The method comprises the following steps: firstly, acquiring instruction information and attributes, RSSI values and historical pairing information of to-be-distributed network equipment, and providing a data basis for configuration; core equipment is screened through instruction and attribute matching and attribute similarity analysis, and equipment strongly related to user requirements is accurately positioned; historical pairing and RSSI value quantization equipment distance weights are fused, association conditions are analyzed, and a network structure is visually presented; analyzing the path length of the core equipment based on a graph structure, dividing closely associated target equipment groups, and determining the influence range of the core equipment; finally, the differentiated dynamic key is dynamically generated in combination with the equipment attribute, the distance weight and the instruction information in the group, the scheme effectively adapts to the dynamic change scene of the equipment through association analysis and dynamic adaptation, and the communication security and specificity are improved.

The invention discloses a narrow space mechanical arm path planning method based on a large language model. According to the method, narrow space path planning is carried out by adopting reinforcement learning, a path multi-scale attribute weighting function is constructed and generated as a reward function of reinforcement learning, and a reward function expression form capable of being directly operated and a reinforcement learning training algorithm matched with the reward function expression form are generated under the guidance of a large language model (LLM). The reinforcement learning training indexes are combined to guide the large language model to optimize the generated reward function, the defects that in traditional reinforcement learning, the reward function is difficult to formulate, and a large amount of experience needs to be provided are overcome, a feasible path can be quickly found, and the generated path is high in quality and directly available; balance of path length, safety and execution precision can be achieved through weight dynamic adjustment, rapid migration from narrow space obstacle avoidance to precise assembly and other different scenes is supported, the path quality and the exploration speed can be considered, and the method is suitable for mechanical arm path planning in the complex and narrow environment.

A self-calibrating hybrid spectrophotometer comprises a receptacle for receiving the cuvette, first and second optodes that are configured to permit light to enter the cuvette, third and fourth optodes that are configured to permit light to leave the cuvette, and a controller that is configured to control delivery of light to one of the first and second optodes. Light entering one of the first and second optodes is received at the third and fourth optodes after having interacted with the turbid sample along different paths having different lengths. Measurements made at the third and fourth optodes in response to having illuminated both the first and second source optodes provide two slopes of optical transmittance as a function of path length. The spectrophotometer uses these slopes to identify a slope that is indicative of the absorption coefficient and the reduced scattering coefficient of the sample.

The invention discloses a multi-target path planning method and device, equipment and a storage medium, and relates to the technical field of traffic path planning, and the method comprises the steps: obtaining solar radiation data and original map data, carrying out the multi-source data fusion based on the solar radiation data and the original map data, and obtaining the radiation road network data; based on the radiation road network data, a total cost function of the path length is constructed through an accumulated cost function of the path length and a heuristic function of the estimated cost, and candidate paths are generated according to the total cost function; on the basis of the radiation road network data, calculating the path average radiation intensity of the candidate path, and further obtaining the energy recovery amount of the candidate path; constructing a dual-target cost function based on the total cost function and the energy recovery amount, adjusting node priorities and screening and updating a non-dominated solution set according to the dual-target cost function, and generating a target candidate path set according to the non-dominated solution set; and determining an optimal balance path in the target candidate path set through multi-dimensional index quantitative evaluation and balance coefficient sorting.

A method and instrument for determining optical density of a solution is disclosed. A flow cell 1 having at least three light paths (4a, 4b, 4c) is provided (100), wherein each light path has a respective predetermined path length, l. Absorbance readings are taken (400), A, of the solution at the at least three light paths (4a, 4b, 4c). For each pair of light paths, a slope, αc, is calculated (500) by dividing a difference in absorbance reading, ΔA, with a difference in path length, Δl. The calculated slopes, αc, are compared (600), and a) if the calculated slopes, αc, are the same, the slope is used for determining (700) optical density of the solution, or b) if the calculated slopes, αc, are not the same, the steepest slope of the calculated slopes is used for determining (701a) optical density of the solution, or the slope of the calculated slopes being in the range of an absorbance reading of 0.01 to 2 is used for determining (701b) optical density of the solution.

The invention relates to the technical field of path planning, in particular to a navigation device path planning method and system, and the method comprises the steps: building a dynamic ocean grid map, dividing an ocean region into dynamic grids, and constructing a multi-factor cost matrix; selecting a path, and constructing a path selection probability formula of the ship; pheromones are updated, for each feasible path, the path length and the corresponding pheromone increment are calculated, the pheromones of the corresponding points on the paths are updated, and the navigation marks are updated after each round of iteration; and updating the dynamic parameters once every preset time step, stopping updating until the maximum number of iterations is reached, and outputting a global optimal path and total cost.

The invention discloses an AGV collaborative dynamic optimization picking and transporting method, and the method comprises the steps: taking the current position of an AGV as a path starting point, taking a cargo storage position as an intermediate node, and taking a target position as an end point, and constructing a multi-constraint path model integrating the path length and steering complexity; solving the multi-constraint path model by adopting an intelligent optimization algorithm, and outputting a global optimal transportation sequence; in the path execution process, the continuous execution cost and the task redistribution cost of the current AGV executing the remaining path are calculated in real time; when the continuous execution cost exceeds the task redistribution cost, generating an AGV task redistribution instruction based on cost comparison; and triggering local task transfer according to the redistribution instruction, and controlling the idle AGV to take over the cargo combination bound with the original AGV and execute a reorganization strategy until the order is completed. The AGV scheduling, loading and path are optimized by fusing position / physical property / state triple dimensions, the interruption rate is reduced through a dual-cost transfer mechanism, and the efficiency and the equipment utilization rate are improved.

The invention relates to the technical field of path planning, in particular to a path planning method based on target characteristics. The motion instability caused by steering sudden change is effectively inhibited by adopting a tightening window and strengthening path cost for a large obstacle; and the local obstacle avoidance flexibility is improved by relaxation constraint of small obstacles. The newly added braking safety verification ensures that no sudden stop risk exists under sudden obstacles, experimental data shows that the scheme synchronously optimizes the path length, task time, path smoothness and safety in a complex scene, meets the core requirement of efficient and stable operation of the commercial scrubber, adopts a tightening strategy for large obstacles, and adopts a tightening strategy for small obstacles. If yes, the constraint is relaxed, the path penalty weight is reduced, more aggressive small-range maneuvering is allowed, and the scene adaptability is remarkably improved through the mechanism; the technical problems that when the scrubber adopts a traditional DWA as a local algorithm to conduct path planning, the track is not smooth, the path is roundabout, and braking safety is insufficient are solved.

The invention belongs to the technical field of robot path planning, and particularly discloses a robot path planning method based on large language model guidance, and the method comprises the steps: discretizing an environment space, and building a configuration space of a robot and an obstacle; calculating a collision area in real time and segmenting a free space into connected components; generating a prompt text; calling a large language model to analyze the text and outputting a recommended obstacle list; constructing a path tree; generating a new node; generating candidate paths based on new node backtracking, and allowing node poses to shift in a preset range through a tolerance verification mechanism; and when the target area is connected with the connected component, adding the path into the solution set library, and selecting an optimal solution according to a weighted cost function of a minimum path length and a total obstacle moving distance. The search strategy is optimized by using the semantic understanding ability of the large language model, the efficiency and adaptability of path planning in a dynamic environment are remarkably improved, and the method is suitable for scenes such as warehouse logistics and disaster rescue.

The invention discloses a reinforcement learning-based cable arrangement method and system in a power supply and distribution system, and a storage medium. The method comprises the following steps of S1, obtaining cable information, laying environment information and constraints required to be met by cable laying; s2, abstracting a bridge system into a weighted undirected topological graph; s3, calling a path planning algorithm based on reinforcement learning, and calculating an optimal cable laying path by taking a minimum comprehensive cost function including path length cost, plot ratio cost and precious penalty terms as an optimization target under the constraint that cable laying needs to be met; s4, calculating the center coordinate of each cable to be arranged in each bridge by adopting a greedy algorithm according to the optimal cable laying path, and arranging the cables according to the center coordinate; according to the method, the optimal laying and arrangement path can be planned on the premise of meeting various physical constraints of cable laying, and the method can automatically and dynamically adapt to different laying scenes.

The invention relates to an improved genetic algorithm-based path planning method for an intelligent robot with a body, which comprises the following steps of: S1, initializing an environment grid map and updating map information, S2, planning a plurality of feasible paths as an initial population by using a multi-tree RRT-connect algorithm, S3, evaluating the advantages and disadvantages of the paths by using a fitness function taking the path length and the path smoothness as evaluation indexes, and S4, determining whether the paths are good or not by using a fitness function taking the path length and the path smoothness as evaluation indexes. The method comprises the following steps: S1, selecting paths, sorting the paths through a fitness function, selecting the paths through an elitist retention strategy and a roulette strategy to carry out crossing, variation and deletion operations, S5, carrying out iteration to obtain a final path, and removing redundant points through a removing strategy, and S6, obtaining the final path. The method solves the problems that a traditional genetic algorithm is low in planning efficiency, tortuous in path, excessive in redundant points and not suitable for continuous space path planning, has the advantages of being high in planning efficiency, good in quality and wide in range, and can be applied to path planning and obstacle avoidance in a complex environment.

A method of monitoring a subject with an opto-physiological sensor system is described herein. The method comprises obtaining a model of the optical properties of at least one body tissue type to be monitored, wherein the model of the optical properties comprises a definition of static (DC) and dynamic (AC) components of transmitted optical power and a definition of a source-detector separation related to a normalised path length for an illumination source of the opto-physiological sensor: obtaining an indication of at least one physiological property of the subject from a wearable device worn by the subject: obtaining an indication of at least one physical variable from the wearable device worn by the subject; determining, using the opto-physiological model, how the at least one physical variable affects the at least one physiological property; and determining a corrected value for the physiological property based on the determination of how the at least one physical variable affects the at least one physiological property.

The invention discloses a crane working path planning method and system and application thereof. The method comprises the following steps: 1) constructing a crane operation demonstration database; 2) pre-training a convolutional neural network by using a demonstration database, and extracting a mapping relation between environment features and tracks; 3) inputting the current starting point position, the target point position and the environment information obtained by the laser radar into a path generation network, and outputting an initial action sequence of the rotation degree of freedom, the dolly motion degree of freedom and the hook degree of freedom; 4) performing local re-planning on the initial action sequence by adopting a fast random tree algorithm based on physical constraint, wherein the optimization criterion is the weighted sum minimization of the path length, the load swing angle and the action sequence variation; and (5) the crane is controlled to move according to the optimized action sequence, the load swing angle is monitored in real time, and emergency braking is triggered when the swing angle exceeds a threshold value. By means of the method, efficient and safe operation of the crane in the complex environment can be achieved, and good engineering practicability and popularization value are achieved.

The invention relates to the technical field of robots, and discloses an autonomous obstacle avoidance navigation method, a robot and a storage medium. According to the autonomous obstacle avoidance navigation method, in the process of moving along an initial planning path, when a first target obstacle and a second target obstacle (within a preset range of the first target obstacle) are sensed, the first target obstacle and the second target obstacle are detected; firstly, a target avoidance point is determined based on parameters such as a passing distance between two obstacles so as to avoid the obstacles and re-plan paths to obtain at least two candidate planning paths, and finally, a target planning path is determined according to cost values and path lengths of the at least two candidate planning paths. Thus, according to the technical scheme, multiple obstacles can be sensed in real time, the target obstacle avoidance point can be accurately determined according to the multiple obstacles, and the obstacle avoidance efficiency can be improved while the obstacle avoidance safety is guaranteed and the collision risk is reduced.

The invention provides an APF-RRT * and genetic algorithm fused unmanned aerial vehicle path planning system and method, and belongs to the technical field of unmanned aerial vehicle autonomous flight. An improved APF-RRT * algorithm is used for carrying out path search, a collision-free initial path from a starting point to a target point is generated, and the improved APF-RRT * algorithm comprises the steps that intelligent sampling fused with a regional probability sampling strategy and a dynamic target bias strategy is adopted to guide growth of a random tree; and adaptively adjusting the extension step length according to the density of obstacles around the current node, the path points of the initial path serve as an initial population of a genetic algorithm, a final path meeting the dynamic constraint of the unmanned aerial vehicle is generated through iterative optimization of the genetic algorithm, and a fitness function of the genetic algorithm considers whether the path length, the collision condition and the path turning angle exceed the constraint or not; according to the method, the planning efficiency and the path quality are improved, the executability of the final path is ensured, and the optimization speed and the path quality are both considered.

The invention discloses an unmanned aerial vehicle path planning method based on comprehensive risk cost, which can generate a low-risk unmanned aerial vehicle flight path. The method comprises the steps that firstly, an unmanned aerial vehicle path planning model containing comprehensive risks and operation cost is constructed, the model comprises a target function and constraint conditions, and the target function is composed of airspace risk cost and unmanned aerial vehicle transportation cost; then, setting three-dimensional constraint conditions including a voyage constraint, a turning angle constraint and a maximum takeoff weight constraint, and constructing a feasible solution space conforming to physical characteristics of the unmanned aerial vehicle; then, performing initial path planning based on a heuristic search weight-A * algorithm with weight, and generating an initial feasible path meeting a constraint condition by dynamically adjusting a heuristic function weight balance path length and risk cost; and finally, based on a parallelogram rule, optimizing the turning point optimization of the initial planning path. According to the invention, a path planning solution with safety and economical efficiency is provided for unmanned aerial vehicle transportation in a complex low-altitude environment.

The invention provides a software defect prediction method based on a dynamic path adaptive graph convolution network, and the method comprises the steps: constructing graph structure data of a software module, inputting the graph structure data into the graph convolution network to learn a dependency relationship between modules, carrying out the adaptive weight distribution of a multi-hop path through a dynamic path weighting convolution layer, and carrying out the prediction of a software defect. The long-path noise interference is suppressed based on the path length attenuation coefficient; optimizing an adjacent matrix in real time based on a multi-head attention mechanism, enhancing a key dependency edge and pruning redundant connection; and embedding a dynamic path score and an adaptive graph updating algorithm into a classification objective function, updating a training weight in combination with a balance strategy, and outputting a defect prediction result through the optimized model. According to the method, the problems of insufficient dependent modeling, noise accumulation and performance degradation in a class imbalance scene of a traditional graph convolutional network under a static graph hypothesis can be effectively solved.