A Deep Learning-Based Intelligent Control Method and System for Laser Driven

Abstract

This invention relates to the field of intelligent control technology, specifically to a laser-driven intelligent control method and system based on deep learning. The method includes the following steps: triggering and parsing the time point of the laser channel; using trajectory and temperature data for path filtering and feature extraction; identifying abrupt change numbers based on trajectory changes and slope direction; determining directional deviations based on trajectory angles; and generating laser-driven intelligent control commands. In this invention, the channel response sequence and initial time point are correlated through an index; trajectory and temperature diffusion distribution constitute a joint feature basis; the slope of displacement changes is used to filter out invalid channels; the difference between the slope direction of the orbital velocity and the temperature change gradient is used to group channels; abrupt trajectories are extracted into a set of numbers using amplification and persistence parameters; the directional angle analysis identifies the path deviation state; directional separation is used to link jump control and path trend data recording; and control commands are matched with number changes in real time, improving path identification accuracy and control response efficiency.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to a laser-driven intelligent control method and system based on deep learning. Background Technology

[0002] The field of intelligent control technology encompasses a comprehensive range of technologies that utilize electronic, computer, and algorithmic methods to perceive, analyze, make decisions, and regulate industrial equipment, systems, and processes. Its core aspects include information acquisition, modeling and analysis, decision control, and execution feedback, and it is widely applied in industrial automation, robot control, energy management, and traffic scheduling. This technological field integrates sensors, communication technologies, computing platforms, and control algorithms to achieve precise control and intelligent response to the operating state of physical systems. It features adaptability, self-learning, and real-time performance, making it a key direction for the upgrading and intelligent transformation of modern control systems. Traditional laser-driven intelligent control methods utilize lasers as a driving source, employing preset logic control programs or simple feedback control systems to regulate the release and spatial orientation of laser energy to a target object. These methods typically rely on fixed model parameters and rule settings, operating and regulating based on physical parameters such as laser pulse width, frequency, and power thresholds. They lack real-time data analysis and dynamic optimization capabilities for regulation strategies, and their response to environmental changes or target states is lag-dependent and limited.

[0003] Traditional laser-driven control methods rely on fixed parameters to construct operating logic, and channel identification depends on preset response sequences. They fail to establish a dynamic mapping mechanism between trigger timing and signal structure. Trajectory response analysis lacks a judgment system linked to the physical state of the channel, making it difficult to select target tracks based on the changing characteristics between paths. Track control does not embed the discrimination dimension of rate change and trend differentiation, path offset direction judgment lacks continuous trend comparison support, and path switching execution lacks a targeted condition activation mechanism. During the control process, there are problems such as number selection error, trend misjudgment, and path response delay, making it difficult to support rapid identification and precise control in multi-path parallel processing or abrupt changes. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a laser-driven intelligent control method based on deep learning, comprising the following steps:

[0005] S1: The path signal channels are triggered sequentially by the laser device, the initial electrical signal time point is analyzed, the time markers are arranged according to the triggering order, the time and channel number are indexed and bound, the order relationship is recorded, and a numbered starting trigger chain is generated.

[0006] S2: Based on the numbered starting trigger chain, extract displacement trajectory data and temperature diffusion pattern, establish time axis distribution interval, determine path correspondence by displacement change slope, eliminate channels with slope below change slope threshold, and generate morphological corresponding track sequence.

[0007] S3: Call the orbital sequence corresponding to the morphology, compare the orbital change rate according to the number order, group according to the consistency of the slope direction of the displacement and temperature change gradient, select the number in combination with the duration of the rising segment and the amplitude change gradient, and generate abrupt channel identification mark.

[0008] S4: Extract the mutation channel identification marker, perform directional fitting on the adjacent number trajectory curve, calculate the angle direction of the up and down segments, identify the trajectory trend difference area, and generate the directional offset judgment result;

[0009] S5: Based on the direction offset determination result, activate the corresponding path jump flag, control the link to execute instructions, record the number change index and path trend graph, and generate laser-driven intelligent control instructions.

[0010] As a further embodiment of the present invention, the numbered starting trigger chain includes a trigger time tag, a channel number index, and a channel sequence relationship; the morphology-corresponding track sequence includes a trajectory displacement change threshold, a trajectory filtering number, and a path matching channel; the mutation channel identification marker includes a change rate comparison result, a trend consistency grouping parameter, and a mutation feature number; the direction offset determination result includes an adjacent number position relationship, a trajectory trend angle direction, and a trend change area identifier; and the laser-driven intelligent control command includes a path jump control bit, a number change record index, and offset trend graphic data.

[0011] As a further aspect of the present invention, the change slope threshold refers to the slope limit for judging the change in path trajectory displacement, and invalid channels that do not meet the change requirements are filtered out.

[0012] As a further aspect of the present invention, the adjacent numbered trajectory curve refers to the displacement and temperature trajectory curves corresponding to adjacent channels in the numbering order, analyzing the path continuity and trend differences.

[0013] As a further aspect of the present invention, the specific steps of S1 are as follows:

[0014] S101: Based on the path signal channel triggered sequentially by the laser device, the initial electrical signal in the channel is detected, the position of the first jump in the trigger signal waveform is monitored, the corresponding time point is recorded as the trigger start time of the channel, and the channel trigger time sequence is generated.

[0015] S102: Call the channel to trigger the time series, sort all time points in chronological order, pair the sorting results with the corresponding channel numbers, record the sorting index value of each number, and generate a channel index pairing matrix;

[0016] S103: Based on the channel index pairing matrix, arrange the channel numbers in order according to the index value, extract the order relationship between adjacent channel numbers, construct the trigger order mapping relationship between numbers, and obtain the number starting trigger chain.

[0017] As a further aspect of the present invention, the specific steps of S2 are as follows:

[0018] S201: Based on the numbered starting trigger chain, extract the instantaneous trajectory change data of the displacement measuring device and the temperature diffusion evolution pattern of the thermal expansion sensing device, retrieve the start and end times of the two on the time axis, construct the corresponding time segments of trajectory and temperature, and generate a set of time distribution intervals of trajectory and temperature.

[0019] S202: Call the trajectory and temperature time distribution interval set, calculate the number of changes in coordinate points per unit time of the trajectory segment, and combine the degree of time overlap of temperature evolution data to determine the correspondence between trajectory trend and path number, and generate trajectory path mapping matrix.

[0020] S203: Based on the trajectory path mapping matrix, filter out the path channel numbers whose trajectory change slope is lower than the preset change slope threshold, retain the number sequence, and obtain the trajectory sequence corresponding to the shape.

[0021] As a further aspect of the present invention, the specific steps of S3 are as follows:

[0022] S301: Call the orbit sequence corresponding to the shape, extract the graphic change rate in numerical order, perform directional difference processing on adjacent orbits, obtain the rate direction change value corresponding to the number, and generate a rate difference direction matrix;

[0023] S302: Based on the rate difference direction matrix, extract the displacement change trend direction and the slope direction of the temperature change gradient of the trajectory channel, calculate the difference between the two directions, and group the channels with a difference less than the consistency threshold to obtain the trend consistency group set.

[0024] S303: Call the trend consistency group set, extract the path rise time and amplitude change gradient of the channel, filter the numbers that simultaneously meet the duration threshold and gradient threshold, establish the label correspondence, and obtain the mutation channel identification label.

[0025] As a further aspect of the present invention, the specific steps of S4 are as follows:

[0026] S401: Based on the set of numbers recorded in the mutation channel identification marker, extract the index position of the number in the sequence, identify the left and right adjacent numbers, call the corresponding trajectory curve data, and aggregate them into continuous channel segments according to the position to obtain a set of adjacent trajectory segments;

[0027] S402: Call the set of adjacent trajectory segments, perform angular direction fitting on the trajectory curves of the up and down segments of the channel segment, calculate the trend angle of the fitted curve in the two directions, and obtain the trend angle distribution table.

[0028] S403: Based on the trend angle distribution table, determine the magnitude of the angle change, identify the trajectory segment whose angle is greater than the direction change threshold, mark the corresponding number as the direction change channel, establish the corresponding record, and obtain the direction offset determination result.

[0029] As a further aspect of the present invention, the specific steps of S5 are as follows:

[0030] S501: Based on the direction separation identifier in the direction offset determination result, extract the corresponding number, activate the path jump control identifier bit associated with the number, and update the identifier activation status to obtain the path jump control set.

[0031] S502: Call the path jump control set, execute the link instruction trigger operation for each number, record the index change of the number in the sequence when the control action takes effect, extract the trend change information of the trigger segment path, and generate a number control graphic dataset;

[0032] S503: Based on the numbered control graphic dataset, combined with the path jump identifier and index change parameter, the control content is encoded into a unified control frame structure and written into the operation instruction to obtain the laser-driven intelligent control instruction.

[0033] A laser-driven intelligent control system based on deep learning, comprising:

[0034] The trigger signal time analysis module is used to implement S1: sequentially trigger the path signal channels through the laser device, analyze the initial electrical signal time point, arrange the time markers according to the triggering order, index and bind the time and channel number, record the sequence relationship, and generate the numbered starting trigger chain;

[0035] The trajectory and temperature analysis module is used to implement S2: Based on the numbered starting trigger chain, extract displacement trajectory data and temperature diffusion pattern, establish time axis distribution interval, determine path correspondence by displacement change slope, eliminate channels with slope lower than the change slope threshold, and generate morphological corresponding trajectory sequence;

[0036] The orbit change rate identification module is used to implement S3: call the orbit sequence corresponding to the morphology, compare the orbit change rate according to the number order, group according to the consistency of the slope direction of the displacement and temperature change gradient, select the number in combination with the duration of the rising segment and the amplitude change gradient, and generate abrupt channel identification mark;

[0037] The trajectory trend difference identification module is used to implement S4: extract the identification mark of the sudden change channel, perform directional fitting on the adjacent numbered trajectory curves, calculate the angle direction of the up and down segments, identify the trajectory trend difference area, and generate the directional offset judgment result;

[0038] The intelligent control command generation module is used to implement S5: based on the direction offset determination result, activate the corresponding path jump flag, control the link to execute commands, record the number change index and path trend graph, and generate laser-driven intelligent control commands.

[0039] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0040] In this invention, the channel response sequence and initial time point are linked by an index to establish a correspondence. The trajectory and temperature diffusion distribution constitute a joint feature basis. The slope of displacement change is used to filter out invalid channels. The slope direction difference comparison between the orbital speed and the temperature change gradient enables channel grouping. The abrupt trajectory extracts a set of numbers through the increase and persistence parameters. The included angle direction analysis identifies the path deviation status. The direction separation identifier is linked to the jump control and path trend data recording. The control command and number change are matched in real time to improve the accuracy of path identification and the efficiency of control response. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of the steps of the present invention;

[0043] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0044] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0045] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0046] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0047] Figure 6 This is a detailed schematic diagram of S5 of the present invention;

[0048] Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0050] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0051] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent.

[0052] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0053] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0054] Please see Figure 1 This invention provides a laser-driven intelligent control method based on deep learning, comprising the following steps:

[0055] S1: The path signal channels are triggered sequentially by the laser device. The initial electrical signal characteristic time point of the channel trigger is analyzed. The time markers are arranged according to the order of the channel signals. The time point and channel number are bound by indexing. The sequential relationship between the numbers is recorded, and the numbered trigger chain is generated.

[0056] S2: Based on the numbered starting trigger chain, extract the instantaneous trajectory change data recorded by the displacement measuring device and the temperature diffusion evolution pattern of the thermal diffusion sensing device, establish the distribution range of trajectory and temperature on the time axis, use the trajectory trend and trajectory change slope to determine the path correspondence, filter out trajectory channels with change slope below the set threshold, and generate the morphological corresponding track sequence.

[0057] S3: Call the corresponding orbit sequence, perform directional differential comparison of the orbit graphic change rate based on the number order, group the trajectory channels according to the consistency parameter of displacement change trend and thermal expansion change slope direction, select the number by combining the duration of the path rise segment and the gradient of amplitude change, and generate abrupt channel identification mark.

[0058] S4: Based on the set of numbers indicated by the mutation channel identification marker, extract the position parameters of the current number in the sequence, perform angular direction fitting on the trajectory curves of the left and right adjacent numbers, identify the trajectory trend difference area by calculating the angle direction between the trend lines of the upward segment and the downward segment, and generate the direction offset judgment result.

[0059] S5: Based on the direction separation identifier in the direction offset determination result, activate the path jump identifier corresponding to the number, transmit the action command through the control link, record the control number change index value and offset path trend graphic data, and generate laser-driven intelligent control command.

[0060] The numbered starting trigger chain includes trigger time label, channel number index, and channel sequence relationship. The morphology corresponding track sequence includes trajectory displacement change threshold, trajectory filtering number, and path matching channel. The abrupt change channel identification mark includes change rate comparison result, trend consistency grouping parameter, and abrupt change feature number. The directional offset judgment result includes adjacent number position relationship, trajectory trend angle direction, and trend change area identifier. The laser-driven intelligent control command includes path jump control bit, number change record index, and offset trend graphic data.

[0061] Please see Figure 2 The specific steps of S1 are as follows:

[0062] S101: Based on the path signal channel triggered sequentially by the laser device, the initial electrical signal in the channel is detected, the position of the first jump in the trigger signal waveform is monitored, the corresponding time point is recorded as the trigger start time of the channel, and the channel trigger time sequence is generated.

[0063] First, the laser device's emitting end is physically connected to several path channels, and each channel is numbered. A sampling trigger mechanism is set for each channel. A control circuit initiates each laser pulse, sequentially acting on each channel. An electrical signal acquisition module is connected to each channel. The acquisition module samples voltage at fixed time intervals, forming a series of time-voltage point data sequences. The first few voltage values ​​in the initial state of each data sequence are read, and their average is calculated as the static baseline voltage for that channel. Subsequently, the sampling sequence is checked point by point to determine if there is a significant change in voltage value relative to the baseline voltage. This change is identified using a jump threshold. For example, if the baseline voltage is 0.05V and the jump threshold is set to 0.3V, then... A transition point is identified when the voltage value at any sampling point is greater than or less than the baseline voltage of 0.05V within ±0.3V. This transition detection operation is performed channel by channel. After the first sampling point that meets the transition condition is detected, the timestamp corresponding to that sampling point is read. This time is the trigger time of that channel. At the same time, the time and the corresponding channel number information are recorded. Based on the sampling analysis results of all channels, a complete dataset containing each channel number and its first trigger time is formed. In this process, in order to improve the accuracy of trigger time recording, the sampling frequency needs to be set to a sufficiently high level, such as above 1MHz, so that the timestamp can be accurate to the microsecond level. Finally, the trigger start time of each channel is obtained, forming a channel trigger time sequence.

[0064] S102: Call the channel to trigger the time series, sort all time points in chronological order, pair the sorting results with the corresponding channel numbers, record the sorting index value of each number, and generate a channel index pairing matrix;

[0065] First, all channels are paired with their corresponding trigger times and channel numbers. The time values ​​of each pair are read and sorted sequentially. This sorting process involves comparing the order of any two trigger times. For example, the first and second pairs are compared sequentially; if the second time is earlier, the two pairs are swapped. This process continues until all pairs are arranged in chronological order. This sorting process can be done manually or programmatically. For instance, if the trigger times for three channels are T1: 120μs, T2: 90μs, and T3: 150μs, the sorted order should be T2, T1, T3. After sorting, each sorted channel number is indexed and assigned a value according to its position in the new order, labeled as channel 1, 2, 3, etc. This process yields the pairing information for each channel number and its sorting position. This pairing information is recorded in a two-dimensional structure, with each item representing a channel number and its corresponding index, such as {(T2, 1), (T1, 2), (T3, 3)}. The result is the channel index pairing matrix, which serves as the foundational data source for constructing the channel order relationship in subsequent steps.

[0066] S103: Based on the channel index pairing matrix, arrange the channel numbers in order according to the index value, extract the order relationship between adjacent channel numbers, construct the trigger order mapping relationship between numbers, and obtain the number starting trigger chain;

[0067] First, rearrange all channel numbers in the matrix according to their corresponding index values ​​in ascending order to obtain a sequence of channel numbers arranged in chronological order. Then, extract two adjacent channel numbers from this sequence as a trigger pair. For example, if the sequence is T2, T1, T3, the trigger pairs are (T2, T1) and (T1, T3). Each pair indicates that the trigger signal of the former was detected before the latter. Establish the trigger order mapping between all channels in this way. Combine all the formed sequence pairs into a set and connect them according to the order of the numbers to form a line. In constructing a trigger chain with a linear structure, it is necessary to check whether each channel has a predecessor and successor node to determine whether the channel is the starting point or a relay node of the chain. By traversing all sequential pairs, the channel number that is in the first position but not in the second position can be identified as the starting channel of the entire trigger chain. For example, if T2 appears in (T2, T1) but not in the second position of any other sequential pair, then T2 can be identified as the starting node. Finally, all trigger pairs are connected in sequence to construct a complete numbered starting trigger chain, which represents the timing relationship between each channel based on the sequential triggering of lasers.

[0068] Please see Figure 3 The specific steps of S2 are as follows:

[0069] S201: Based on the numbered starting trigger chain, extract the instantaneous trajectory change data of the displacement measuring device and the temperature diffusion evolution pattern of the thermal expansion sensing device, retrieve the start and end times of the two on the time axis, construct the corresponding time segments of trajectory and temperature, and generate a set of time distribution intervals of trajectory and temperature.

[0070] First, the trigger time points for each path channel need to be extracted from the trigger chain. Then, these trigger times are combined with measurement data from the displacement measuring device and the thermal expansion sensor. The displacement measuring device records the instantaneous displacement changes at each point on the path. By tracking the displacement of objects on the path through sensing devices, the displacement measuring device records the change value whenever an object on the path undergoes a displacement change, generating a displacement data sequence synchronized with time. Simultaneously, the thermal expansion sensor monitors temperature changes along the path, collecting temperature evolution data and recording the temperature distribution at each time point. This data is synchronized with the trajectory data, facilitating subsequent analysis. During execution, the start and end times of these two types of data need to be compared to ensure that the time periods corresponding to the trajectory data and temperature data are consistent. Assuming the time period for the trajectory data is [0s, 10s] and the time period for the temperature data is [1s, 9s], the intersection of these two data points on the time axis needs to be found, i.e., [1s, 9s]. This allows for the comparison and analysis of the trajectory and temperature change data within the same time interval. Ultimately, by dividing the intersection of trajectory and temperature data into several smaller time intervals, the correspondence between trajectory and temperature changes can be clearly obtained, generating a set of trajectory and temperature time distribution intervals, thus ensuring the accuracy and completeness of the data processing process.

[0071] S202: Call the trajectory and temperature time distribution interval set, calculate the number of changes in coordinate points per unit time of the trajectory segment, and combine the time overlap of temperature evolution data to determine the correspondence between the trajectory trend and the path number, and generate the trajectory path mapping matrix.

[0072] First, the start and end times of each trajectory segment are extracted. Then, the trajectory data is refined based on these time segments, and the displacement change per unit time within each segment is calculated. For example, if the trajectory measuring device records a displacement change of 20 meters in 5 seconds within a trajectory segment, then the displacement change per unit time is 4 meters / second. Next, the unit time change of each trajectory segment is compared with the corresponding temperature evolution data to see if there are significant changes in the two data within the same time interval. If the time overlap between the trajectory and temperature changes is high, such as exceeding 80%, then the relationship between the two segments is considered strong. For example, if the time overlap between the trajectory segment and the temperature segment is 85%, it means that the patterns of trajectory change and temperature change are synchronous within the same time period. Therefore, when generating the trajectory path mapping matrix, the two segments can be considered to have a strong correlation. In the mapping matrix, each row represents a trajectory segment, each column represents a path number, and each element value in the matrix represents the correspondence between the trajectory segment and the path number, reflecting the strength of their correlation.

[0073] S203: Based on the trajectory path mapping matrix, filter out the path channel numbers whose trajectory change slope is lower than the preset change slope threshold, retain the number sequence, and obtain the trajectory sequence corresponding to the shape;

[0074] First, the relationship between each trajectory segment and path number is extracted from the matrix. Next, for each path number, the rate of change of its corresponding trajectory segment per unit time is analyzed. The trajectory change slope reflects the correlation between the path number and the trajectory change; specifically, if the trajectory corresponding to a path number changes drastically within a certain time period, the trajectory change slope will be high. To filter out path numbers with significant changes, a threshold for the change slope needs to be set, such as 0.1 m / s. If the trajectory change slope of a path number is lower than this threshold, the trajectory change of that path is considered insignificant and should be removed from the matrix. In this way, path numbers with small changes and indistinct trajectory trends can be removed, retaining only those with significant change trends. Thus, the final result is a sequence of path numbers containing only path numbers with significant trajectory changes and trajectory change slopes higher than the threshold, indicating that the trajectory change trends of these paths are significant within a given time period.

[0075] Please see Figure 4 The specific steps of S3 are as follows:

[0076] S301: Call the orbital sequence corresponding to the shape, extract the rate of change of the shape in numerical order, perform directional difference processing on adjacent orbits, obtain the rate direction change value corresponding to the number, and generate a rate difference direction matrix;

[0077] First, the rate of change of each track is extracted in numerical order. This rate is calculated by measuring the displacement of each track point within a unit of time, reflecting the rate of change of the track's motion. For example, if a track moves 10 meters in 5 seconds, its rate is 2 meters per second. Next, directional differencing is performed on adjacent tracks. Directional differencing involves comparing the rates of change of two adjacent tracks, calculating the rate difference between them, and analyzing the direction of this difference. For instance, if track A's rate of change is 2 meters per second and track B's rate of change is 3 meters per second, the difference is 1 meter per second, and the direction of change for track A is positive while that for track B is negative. In this case, the change in the direction of the rate difference is negative. By performing this differencing operation on each pair of adjacent tracks, the rate direction change values ​​between all tracks are obtained, ultimately generating a rate difference direction matrix. This matrix can be used to reflect the differences in the rates of change between tracks, as well as the directional changes of the tracks in the time series, providing necessary data support for subsequent track analysis.

[0078] S302: Based on the rate difference direction matrix, extract the displacement change trend direction and the slope direction of the temperature change gradient of the trajectory channel, calculate the difference between the two directions, and group the channels with a difference less than the consistency threshold to obtain the trend consistency group set.

[0079] First, the direction of displacement change trend for each track channel is extracted. This direction refers to the direction of displacement change of an object on the track within a certain time interval. For example, when the track displacement increases over time, the direction of displacement change trend is positive; when the displacement decreases over time, the direction of change trend is negative. Next, the slope direction of the temperature change gradient is extracted. The slope direction of the temperature change gradient reflects the inclination of temperature change along the path over time, which can be determined by the trend of temperature data. For example, if the temperature changes with time, the slope direction of the temperature change gradient is positive; if the temperature decreases over time, the direction is negative. The direction of displacement change trend of the track channel is compared with the slope direction of the temperature change gradient, and the difference between their directions is calculated. If the directions of change of the track and temperature are the same, the difference is small; if the directions are opposite, the difference is large. For example, if the displacement change trend of the track is upward, and the temperature change trend within the corresponding time period is also upward, then the two trends can be considered to be the same; if the temperature shows a downward trend, then the two trends can be considered to be opposite. Based on the criteria for judging trend consistency (such as whether the signs of change direction are the same), the track numbers with consistent trends are grouped together to form a trend consistency group set.

[0080] S303: Call the trend consistency group set, extract the path rise time and amplitude change gradient of the channel, filter the numbers that simultaneously meet the duration threshold and gradient threshold, establish the label correspondence, and obtain the mutation channel identification label;

[0081] First, the path ascent time and amplitude gradient are extracted for each channel. The path ascent time refers to the period during which the track transitions from a lower to a higher state of change. Typically, during this phase, the track's displacement changes rapidly, demonstrating significant changes. The amplitude gradient refers to the amount of displacement change per unit time within the path ascent time. For example, if the track displaces 15 meters in 3 seconds, the amplitude gradient is 5 meters per second. Next, duration and gradient thresholds need to be set. The duration threshold is the minimum length of the path ascent time; only tracks exceeding this threshold are considered valid ascent segments. Let's assume the duration threshold is set to 2 seconds. The gradient threshold is the minimum displacement change per unit time within the track's ascent segment; only tracks with an amplitude gradient greater than this threshold are further analyzed. Let's assume the gradient threshold is set to 4 meters per second. By filtering the path ascent time and amplitude gradient for each track, track numbers that simultaneously meet both the duration and gradient thresholds are identified. For example, if a track's path ascent time is 3 seconds and its amplitude gradient is 6 meters per second, then this track number meets the criteria and enters the marking process. Finally, the orbital numbers that meet the criteria are labeled as mutation channels, and a label correspondence is established to obtain mutation channel identification labels. The orbital numbers in the labels indicate that these orbitals have undergone significant changes within the time period and belong to mutation orbitals.

[0082] Please see Figure 5 The specific steps of S4 are as follows:

[0083] S401: Based on the set of numbers recorded in the mutation channel identification marker, extract the index position of the number in the sequence, identify the left and right adjacent numbers, call the corresponding trajectory curve data, and aggregate them into continuous channel segments according to position to obtain the set of adjacent trajectory segments;

[0084] First, the index positions of these numbers in the track sequence are extracted. This process is done by iterating through each number in the tag set to find its specific position in the track data sequence, thus obtaining its exact index value in the sequence. For example, if number T3 is located at index position 5 in the track data sequence, then T3 is the number at that position. Next, the adjacent track numbers are identified, meaning that the left-hand number T2 and right-hand number T4 of number T3 need to be found to ensure the clear order of the track numbers. Next, the trajectory curve data corresponding to these track numbers is retrieved. The trajectory curve data records the positional information of objects on the track over time, reflecting the displacement of each point on the track at different times. Based on this, the data is processed according to the order of the track numbers to ensure smooth track connections and reflect the correct physical positional relationships. The trajectory data of adjacent tracks are aggregated to ensure seamless connection between each track segment and adjacent track segments, forming a continuous track channel segment. These continuous channel segments help analyze the connectivity between tracks, the motion path of objects on the tracks, and ultimately generate a set of adjacent trajectory segments containing all continuous track segments, which can be used for subsequent trajectory analysis and modeling.

[0085] S402: Call the adjacent trajectory segment set, perform angular direction fitting on the trajectory curves of the up and down segments of the channel segment, calculate the trend angle of the fitted curve in the two directions, and obtain the trend angle distribution table.

[0086] First, the angular directionality of the trajectory curves for the upward and downward segments of each channel segment is fitted. The upward segment refers to the portion of the track where the displacement of the object increases with time, while the downward segment is the portion where the displacement of the object decreases with time. These reflect different directional changes in the track. To analyze these changes in depth, the trend angle of each segment in these two directions needs to be determined by fitting the trajectory curves of the upward and downward segments. Specifically, the fitting process analyzes the time-position data of the trajectory, calculates the slope of the trajectory, and then deduces the angle of each segment. For example, if the displacement of the object increases at a rate of 5 meters per second in the upward segment, and decreases at a rate of 3 meters per second in the downward segment, the trend angle of each segment can be further derived by calculating the rate of change of velocity in each segment and combining it with the geometric distribution of the trajectory direction. After the fitting is completed, the difference in trend angle between the upward and downward segments of each channel segment is calculated, ultimately resulting in a trend angle distribution table. This table will contain the angle changes of each trajectory segment in the upward and downward directions, thus providing detailed angle data for subsequent trajectory analysis and helping to reveal the motion trends and turning points between trajectories.

[0087] S403: Based on the trend angle distribution table, determine the magnitude of the angle change, identify the trajectory segment whose angle is greater than the direction change threshold, mark the corresponding number as the direction change channel, establish the corresponding record, and obtain the direction offset determination result.

[0088] First, the magnitude of the change in the included angle is assessed. The magnitude of the change in the included angle refers to the angular difference between the ascending and descending segments, used to measure whether a significant change in track direction has occurred. If the magnitude of the change in the included angle exceeds a set threshold for abrupt directional changes, it indicates an abnormal track change and a significant change in direction. For example, if the threshold for abrupt directional changes is set at 30 degrees, and the change in the included angle between the ascending and descending segments of a track is greater than 30 degrees, the track is considered to have undergone abrupt changes, potentially indicating a sharp turn in the path. Next, all track segments with included angle changes greater than this threshold need to be identified and marked as directional change channels. This process involves dynamically monitoring track changes and marking track numbers that have undergone directional changes within a certain period. After marking, a record is created documenting which track numbers have changed angles exceeding the set threshold, thus identifying them as tracks with directional changes. Finally, using this data, a directional deviation determination result is obtained. This result clearly identifies significant directional change points on the track, aiding in the formulation of subsequent track monitoring and adjustment strategies.

[0089] Please see Figure 6 The specific steps of S5 are as follows:

[0090] S501: Based on the direction separation identifier in the direction offset determination result, extract the corresponding number, activate the path jump control identifier bit associated with the number, and update the identifier activation status to obtain the path jump control set.

[0091] First, the set of numbers corresponding to these direction separation markers is extracted. This set of numbers contains all track numbers that have undergone significant direction changes, typically indicating abrupt changes or deviations in the track at certain stages. By extracting these numbers, tracks requiring further control and adjustment can be accurately identified. Next, for these numbers, the associated path jump control flag is activated. This flag instructs the control system to perform a jump operation on these tracks. Specifically, whenever a track number is identified as needing a direction change, its corresponding path jump control flag is activated. This process is handled by the system's automated control module, which automatically updates the flag. Each time a flag is activated, the system checks and updates its status to ensure the path jump command is executed at the correct time. For example, assuming track number T5 has experienced a direction shift, the system activates the jump control flag associated with T5 and changes its status from inactive to active. Finally, the jump control flags corresponding to all identified track numbers are activated and updated, forming a path jump control set. This control set summarizes all track numbers requiring jumps and their corresponding control flags, facilitating subsequent processing and operation.

[0092] S502: Call the path jump control set, execute the link instruction trigger operation for each number, record the index change of the number in the sequence when the control action takes effect, extract the trend change information of the trigger segment path, and generate the number control graphic dataset;

[0093] The system begins triggering link commands for each number. The core of this operation is to trigger the corresponding link command based on the number in the path jump control set, ensuring that the selection and execution of the jump path meet predetermined requirements. During execution, the system locates the corresponding track path based on the number in the path jump control set and triggers the associated control command. Whenever a control command is triggered, the system records the index change of that number in the track sequence to ensure the accuracy of the track order and jump operation. For example, if track number T5 moves from position 5 to position 8 in the sequence, the system records this index change. Simultaneously, the system extracts trajectory curve data from the jump segment and analyzes the trend change information of these path segments. This trend change information reflects the dynamic changes of the track in the jump path, such as the rate of displacement change and direction change. By analyzing this trend change information, the system can determine whether the path jump meets the predetermined control requirements and provide feedback on the control results. Finally, after all path jump operations are completed, the system will generate a numbered control graphical dataset. This dataset includes detailed data on control command triggers, index changes, and trajectory trend changes corresponding to each track number, providing necessary graphical data support for subsequent control command execution.

[0094] S503: Based on the numbered control graphic dataset, combined with path jump identifiers and index change parameters, the control content is encoded into a unified control frame structure and written into the operation instructions to obtain laser-driven intelligent control instructions.

[0095] Combining path jump identifiers and index change parameters, the control content is encoded and converted. Path jump identifiers indicate which track numbers require path jumps, while index change parameters reflect changes in track position. Using these two parameters, the system converts the control content into a unified control frame structure. This unified control frame structure is a standardized format used to describe the specific content of control instructions and ensure the accuracy and consistency of information when transmitted and executed between different systems. The conversion process involves encoding the control content into a series of standardized instructions, ensuring that these instructions can be correctly identified and executed by the intelligent control system. For example, if track number T5 changes position, the control content will be converted into a control frame containing information such as track number, jump instruction, starting position, and target position. Finally, all encoded control instructions are written into the operation instructions, forming complete laser-driven intelligent control instructions. These instructions are used to control the precise adjustment and path adjustment of the laser equipment, ensuring accurate execution of laser operations in dynamic environments.

[0096] Please see Figure 7 A laser-driven intelligent control system based on deep learning, comprising:

[0097] The trigger signal time analysis module is used to implement S1: sequentially trigger the path signal channels through the laser device, analyze the initial electrical signal time point, arrange the time markers according to the triggering order, index and bind the time and channel number, record the sequence relationship, and generate the numbered starting trigger chain;

[0098] The trajectory and temperature analysis module is used to implement S2: based on the numbered starting trigger chain, it extracts displacement trajectory data and temperature diffusion patterns, establishes a time axis distribution interval, determines the path correspondence by the slope of displacement change, eliminates channels with slopes lower than the slope threshold, and generates a trajectory sequence corresponding to the shape.

[0099] The orbit change rate identification module is used to implement S3: call the orbit sequence corresponding to the shape, compare the orbit change rate in order of number, group according to the consistency of the slope direction of displacement and temperature change gradient, select the number in combination with the duration of the rising segment and the gradient of amplitude change, and generate abrupt channel identification mark.

[0100] The trajectory trend difference identification module is used to implement S4: extract the identification marker of the sudden change channel, perform directional fitting on the adjacent numbered trajectory curves, calculate the angle direction of the up and down segments, identify the trajectory trend difference area, and generate the directional offset judgment result.

[0101] The intelligent control command generation module is used to implement S5: based on the direction offset determination result, activate the corresponding path jump flag, control the link to execute commands, record the number change index and path trend graph, and generate laser-driven intelligent control commands.

[0102] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A deep learning-based intelligent control method for laser driving, characterized in that, Includes the following steps: S1: The path signal channels are triggered sequentially by the laser device, the initial electrical signal time point is analyzed, the time markers are arranged according to the triggering order, the time and channel number are indexed and bound, the order relationship is recorded, and a numbered starting trigger chain is generated. S2: Based on the numbered starting trigger chain, extract displacement trajectory data and temperature diffusion pattern, establish time axis distribution interval, determine path correspondence by displacement change slope, eliminate channels with slope below change slope threshold, and generate morphological corresponding track sequence. S3: Call the orbital sequence corresponding to the morphology, compare the orbital change rate according to the number order, group according to the consistency of the slope direction of the displacement and temperature change gradient, select the number in combination with the duration of the rising segment and the amplitude change gradient, and generate abrupt channel identification mark. S4: Extract the mutation channel identification marker, perform directional fitting on the adjacent number trajectory curve, calculate the angle direction of the up and down segments, identify the trajectory trend difference area, and generate the directional offset judgment result; S5: Based on the direction offset determination result, activate the corresponding path jump flag, control the link to execute instructions, record the number change index and path trend graph, and generate laser-driven intelligent control instructions; The specific steps of S5 are as follows: S501: Based on the direction separation identifier in the direction offset determination result, extract the corresponding number, activate the path jump control identifier bit associated with the number, and update the identifier activation status to obtain the path jump control set. S502: Call the path jump control set, execute the link instruction trigger operation for each number, record the index change of the number in the sequence when the control action takes effect, extract the trend change information of the trigger segment path, and generate a number control graphic dataset; S503: Based on the numbered control graphic dataset, combined with the path jump identifier and index change parameter, the control content is encoded into a unified control frame structure and written into the operation instruction to obtain the laser-driven intelligent control instruction.

2. The laser-driven intelligent control method based on deep learning according to claim 1, characterized in that, The numbered starting trigger chain includes a trigger time tag, a channel number index, and a channel sequence relationship. The morphology-corresponding track sequence includes a trajectory displacement change threshold, a trajectory filtering number, and a path matching channel. The abrupt change channel identification marker includes a change rate comparison result, a trend consistency grouping parameter, and an abrupt change feature number. The directional offset determination result includes the adjacent number position relationship, the trajectory trend angle direction, and the trend change area identifier. The laser-driven intelligent control command includes a path jump control bit, a number change record index, and offset trend graphic data.

3. The laser-driven intelligent control method based on deep learning according to claim 1, characterized in that, The slope threshold refers to the slope limit for judging the change in path trajectory displacement, and invalid channels that do not meet the change requirement are filtered out.

4. The laser-driven intelligent control method based on deep learning according to claim 1, characterized in that, The adjacent numbered trajectory curves refer to the displacement and temperature trajectory curves corresponding to adjacent channels in the numbering sequence, which analyze the continuity of the path and the differences in trends.

5. The laser-driven intelligent control method based on deep learning according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Based on the path signal channel triggered sequentially by the laser device, the initial electrical signal in the channel is detected, the position of the first jump in the trigger signal waveform is monitored, the corresponding time point is recorded as the trigger start time of the channel, and the channel trigger time sequence is generated. S102: Call the channel to trigger the time series, sort all time points in chronological order, pair the sorting results with the corresponding channel numbers, record the sorting index value of each number, and generate a channel index pairing matrix; S103: Based on the channel index pairing matrix, arrange the channel numbers in order according to the index value, extract the order relationship between adjacent channel numbers, construct the trigger order mapping relationship between numbers, and obtain the number starting trigger chain.

6. The laser-driven intelligent control method based on deep learning according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the numbered starting trigger chain, extract the instantaneous trajectory change data of the displacement measuring device and the temperature diffusion evolution pattern of the thermal expansion sensing device, retrieve the start and end times of the two on the time axis, construct the corresponding time segments of trajectory and temperature, and generate a set of time distribution intervals of trajectory and temperature. S202: Call the trajectory and temperature time distribution interval set, calculate the number of changes in coordinate points per unit time of the trajectory segment, and combine the degree of time overlap of temperature evolution data to determine the correspondence between trajectory trend and path number, and generate trajectory path mapping matrix. S203: Based on the trajectory path mapping matrix, filter out the path channel numbers whose trajectory change slope is lower than the preset change slope threshold, retain the number sequence, and obtain the trajectory sequence corresponding to the shape.

7. The laser-driven intelligent control method based on deep learning according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Call the orbit sequence corresponding to the shape, extract the graphic change rate in numerical order, perform directional difference processing on adjacent orbits, obtain the rate direction change value corresponding to the number, and generate a rate difference direction matrix; S302: Based on the rate difference direction matrix, extract the displacement change trend direction and the slope direction of the temperature change gradient of the trajectory channel, calculate the difference between the two directions, and group the channels with a difference less than the consistency threshold to obtain the trend consistency group set. S303: Call the trend consistency group set, extract the path rise time and amplitude change gradient of the channel, filter the numbers that simultaneously meet the duration threshold and gradient threshold, establish the label correspondence, and obtain the mutation channel identification label.

8. The laser-driven intelligent control method based on deep learning according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the set of numbers recorded in the mutation channel identification marker, extract the index position of the number in the sequence, identify the left and right adjacent numbers, call the corresponding trajectory curve data, and aggregate them into continuous channel segments according to the position to obtain a set of adjacent trajectory segments; S402: Call the set of adjacent trajectory segments, perform angular direction fitting on the trajectory curves of the up and down segments of the channel segment, calculate the trend angle of the fitted curve in the two directions, and obtain the trend angle distribution table. S403: Based on the trend angle distribution table, determine the magnitude of the angle change, identify the trajectory segment whose angle is greater than the direction change threshold, mark the corresponding number as the direction change channel, establish the corresponding record, and obtain the direction offset determination result.

9. A laser-driven intelligent control system based on deep learning, characterized in that, The system is used to implement the laser-driven intelligent control method based on deep learning as described in any one of claims 1-8, and the system comprises: The trigger signal time analysis module is used to implement S1: sequentially trigger the path signal channels through the laser device, analyze the initial electrical signal time point, arrange the time markers according to the triggering order, index and bind the time and channel number, record the sequence relationship, and generate the numbered starting trigger chain; The trajectory and temperature analysis module is used to implement S2: Based on the numbered starting trigger chain, extract displacement trajectory data and temperature diffusion pattern, establish time axis distribution interval, determine path correspondence by displacement change slope, eliminate channels with slope lower than the change slope threshold, and generate morphological corresponding trajectory sequence; The orbit change rate identification module is used to implement S3: call the orbit sequence corresponding to the morphology, compare the orbit change rate according to the number order, group according to the consistency of the slope direction of the displacement and temperature change gradient, select the number in combination with the duration of the rising segment and the amplitude change gradient, and generate abrupt channel identification mark; The trajectory trend difference identification module is used to implement S4: extract the identification mark of the sudden change channel, perform directional fitting on the adjacent numbered trajectory curves, calculate the angle direction of the up and down segments, identify the trajectory trend difference area, and generate the directional offset judgment result; The intelligent control command generation module is used to implement S5: based on the direction offset determination result, activate the corresponding path jump flag, control the link to execute commands, record the number change index and path trend graph, and generate laser-driven intelligent control commands.

Images