Multi-channel light source controller and control method thereof

Abstract

The invention provides a multi-channel light source controller and a control method thereof, and the controller comprises a main microcontroller, a plurality of independent constant-current driving channels, and an advanced time sequence control unit which is independent of the main microcontroller on the physical level. An input interface of the light source controller is configured to be a configuration instruction interface and a hard real-time trigger interface on the hardware level, and instructions with different real-time performance and complexity are processed respectively. The main microcontroller is integrated with a multi-channel cooperative control unit, a light source characteristic model unit, a channel coupling compensation unit and an intelligent safety protection unit; the advanced time sequence control unit is configured to generate corresponding driving signals according to different interface instructions. Through the structure, the problems of control lag, channel interference, inaccurate model, safety rigidity and the like in the prior art are fundamentally solved, and high-speed, high-precision and high-reliability multi-channel light source control is realized.

Description

Technical Field

[0001] This invention relates to the fields of machine vision and industrial automation technology, and in particular to a multi-channel light source controller and its control method. Background Technology

[0002] In the fields of machine vision and industrial automation, multi-channel LED light source controllers are key devices for achieving high-quality lighting. Existing controllers generally employ a microprocessor-based architecture with multiple constant current drives, adjusting the brightness of each channel through PWM or DAC technology. However, this approach has the following inherent drawbacks: closed-loop feedback-based control modes suffer from response delays, making them unsuitable for high-speed applications; they neglect the coupling effects between multiple channels, leading to inter-channel interference; they use simplified linear models to describe the complex nonlinear photoelectric characteristics of LEDs, resulting in limited control accuracy; and static safety protection mechanisms restrict the full utilization of hardware performance. These factors collectively limit the performance of existing controllers in high-speed, high-precision application scenarios.

[0003] Therefore, there is an urgent need in this field for a multi-channel light source controller that can achieve high speed, high precision, high synchronization, high security, and adaptive capabilities. Summary of the Invention

[0004] This application provides a multi-channel light source controller and its control method. This controller addresses technical shortcomings such as slow response speed, independent control of each channel, poor synchronization, poor control precision, difficulty in implementing complex multi-angle, multi-spectral synchronous triggering lighting schemes, and significant limitations in application. It aims to create a high-speed, high-precision, highly synchronized, highly secure, and adaptive multi-channel light source controller.

[0005] In one aspect, a multi-channel light source controller is provided, comprising: a main microcontroller (MCU) and multiple independent... The light source controller comprises eight independent constant current drive channels (in this application, the number of such independent constant current drive channels is eight), and a timing control unit physically independent of the main microcontroller. The light source controller's input interface is configured in two hardware configurations: a configuration instruction interface, connected to the main microcontroller, for receiving non-real-time or soft real-time upper-layer instructions; and a hard real-time trigger interface, directly connected to the timing control unit, for receiving external synchronization signals requiring nanosecond or microsecond-level time determinism. The main microcontroller integrates: The multi-channel collaborative control unit is used to uniformly plan an optimal path that strictly coordinates the entire dynamic trajectory from the starting point to the ending point when multiple channels need to change synchronously based on upper-level instructions, and to generate instantaneous target values ​​that are coordinated in time. The light source characteristic model unit is used to establish a dynamic model describing the photoelectric and thermal characteristics of the light source load for each channel, and to solve the instantaneous target value into a driving command based on the model. The channel coupling compensation unit is used to predict the driving command based on a pre-established multi-input multi-output transfer function model describing the physical coupling relationship between channels, and to generate a feedforward compensation signal to suppress mutual interference between channels. The intelligent safety protection unit is used to predict the execution consequences of the compensated driving command based on a multi-dimensional safety boundary function, and to minimize the correction of the command when it may exceed the safety area, so as to generate a safe final driving command. The advanced timing control unit is configured as follows: The system receives the final drive command from the main microcontroller and generates high-precision, low-jitter timing drive signals. The system receives external synchronization signals from the hard real-time trigger interface and converts these signals into high-precision, low-jitter timing drive signals independently of the main microcontroller.

[0006] Furthermore, the multi-channel collaborative control unit includes: a synchronization task instruction parser for parsing instructions containing target values, durations, and synchronization modes; a joint state space and trajectory planner for planning a smooth trajectory from the current state to the target state in a joint state space composed of the channels participating in synchronization; a real-time interpolator for sampling on the trajectory to generate an instantaneous target value sequence for each control cycle; and an instruction distribution module for distributing the instantaneous target values ​​to the underlying controllers of each channel.

[0007] Furthermore, the light source characteristic model unit is configured to: inversely calculate the user-defined target optical output into a corresponding drive current command by iteratively solving a dynamic system model. This unit includes: a model parameter repository for pre-storing a set of model parameters obtained through offline calibration and knowledge distillation for each supported LED model; a real-time state observer for maintaining a current state vector for each channel in real time, iteratively updating the internal state by real-time acquisition of the actual output current of the drive circuit and the LED substrate temperature, combined with the internal thermal model and aging model; a forward predictor for receiving a hypothetical drive current value and calculating the predicted optical parameters that the LED will produce under that hypothetical current; and an inverse control command generator for iteratively calling the forward predictor to solve for the drive current value that makes the predicted optical output match the user-defined target.

[0008] Furthermore, the channel coupling compensation unit includes: a multi-input multi-output coupling interference transfer function model, which is obtained through system identification during the controller design or production calibration stage, and is used to describe the voltage disturbance caused by the current change of any channel to the other channels; a drive instruction preprocessor, used to calculate the expected current change rate of the drive instruction for each channel; a real-time disturbance predictor, used to predict the equivalent voltage disturbance that all channels will experience based on the transfer function model and the expected current change rate; and a compensation signal injection module, used to calculate the feedforward compensation value that needs to be applied to the control loop of each channel according to the predicted voltage disturbance and the characteristics of the constant current drive circuit, and synthesize the compensation value with the original drive instruction to generate the anti-interference final drive instruction.

[0009] Furthermore, the intelligent safety protection unit includes: a multi-dimensional safety boundary function definition module, used to define a safety region composed of multiple state variables such as current, voltage, temperature, and power change rate; a system state prediction module, used to predict the future system state after executing the compensated drive command based on the light source characteristic model unit; a hazard assessment module, which, when it is predicted that the future system state may exceed the safety region, initiates an optimization process to calculate a correction amount that minimizes the change to the original command; and an online optimizer, which, when there is a risk, solves an optimization problem with the goal of minimizing the command correction amount and the constraint of satisfying the safety boundary, and outputs the corrected safety command.

[0010] Furthermore, this advanced timing control unit is a Field Programmable Gate Array (FPGA) or Complex Programmable Logic Device (CPLD), which internally includes: an input signal conditioning and arbitration module for electrically isolating external trigger signals through optocouplers, and filtering out high-frequency noise and mechanical jitter of the input signal through a hardware-implemented digital low-pass filter or debouncing logic, and capturing the high-precision timestamp of each valid trigger signal edge using a high-frequency internal clock; a timing instruction decoding and parameter register group for receiving and storing timing task configurations from the main microcontroller; an atomic timing operation library that solidifies the basic, indivisible atomic timing operation logic; and a high-precision timing sequence generator that, based on the atomic timing operation library and configuration parameters, autonomously and accurately generates the final underlying drive control waveform using a high-frequency internal clock and multiple high-precision counters.

[0011] Furthermore, the light source controller is also equipped with a complex task execution engine, which is used to receive and parse non-real-time or software-defined signals. The main microcontroller also includes a complex task execution engine for receiving and parsing non-real-time or soft real-time macro instructions from the upper layer. Based on its internally stored task knowledge base, the engine autonomously decomposes these macro instructions into execution strategies. These execution strategies include: for tasks requiring multi-channel dynamic coordination, the multi-channel coordination control unit is invoked based on the generated coordination task instructions. This unit generates the coordination trajectory and instantaneous target, and then, through the collaborative efforts of the light source characteristic model unit, channel coupling compensation unit, and intelligent safety protection unit, the final drive instruction is output to the advanced timing control unit. For basic, independent operations, the advanced timing control unit is directly scheduled based on the decomposed atomic operation sequence to efficiently and accurately execute the entire lighting sequence autonomously.

[0012] Furthermore, the complex task execution engine includes: a task knowledge base storing the definitions of multiple complex lighting tasks and their decomposition into atomic operations scripts; an instruction parser and decomposer for receiving macro instructions, retrieving corresponding tasks from the knowledge base, and generating an atomic operation queue; an atomic lighting operation library containing optimized low-level control functions for the real-time scheduler to call; and a real-time scheduler for sequentially retrieving and executing instructions in the atomic operation queue.

[0013] Furthermore, the main microcontroller is also equipped with an online self-calibration module, used to incorporate feedback signals from an external vision system. The predicted values ​​are compared with those of the internal light source characteristic model unit, and the model parameters are automatically updated through optimization algorithms to compensate for the performance drift of the light source load caused by aging or environmental changes.

[0014] The online self-calibration module includes: an external feedback interface for receiving image quality feedback signals from the vision system; an exploratory signal generator for applying small driving perturbations near a stable operating point; an internal model-based predictor for predicting the optical / visual feedback changes that the perturbations should cause; and a model optimizer for calculating the error between the predicted changes and the actual feedback changes, and updating the parameters of the light source characteristic model unit using gradient descent.

[0015] On the other hand, this application also provides a control method for a light source controller.

[0016] Method 1, the control steps include: S1, Instruction reception: If a non-real-time or soft real-time upper-layer instruction is received through the configuration instruction interface, proceed to steps S2 to S7. If an external synchronization signal requiring nanosecond or microsecond time determinism is received through the hard real-time trigger interface, proceed to steps S6 to S7. S2 executes multi-channel collaborative control and plans dynamic collaborative trajectories between channels; S3, based on the dynamic photoelectric and thermal characteristic model of the light source, solves the optical target in reverse to obtain the driving command; S4, perform feedforward compensation on the drive command based on the inter-channel coupling interference model; S5 performs safety verification and minimization correction on the compensated driving instructions based on multi-dimensional safety boundary functions, and generates safe final driving instructions. S6, based on the advanced timing control unit, generates high-precision, low-jitter timing drive signals according to the final drive command or directly received external synchronization signals; S7, using the generated timing drive signal to control multiple independent constant current drive channels.

[0017] Method 2, based on the function of the complex task control engine, includes the following control steps: S11 receives non-real-time or soft real-time upper-layer commands through the configuration command interface; S12, the complex task execution engine converts the received macro instructions into two types of lower-level instructions. For task instructions that require multi-channel dynamic coordination, steps S13 to S18 are executed; for basic, independent operation instructions, steps S17 to S18 are executed. S13, executes multi-channel collaborative control, and plans dynamic collaborative trajectories between channels; S14, based on the dynamic photoelectric and thermal characteristic model of the light source, solves the optical target in reverse into driving commands; S15, perform feedforward compensation on the drive command based on the inter-channel coupling interference model; S16, Based on the multi-dimensional safety boundary function, the safety verification and minimization correction of the compensated driving instructions are performed to generate safe final driving instructions; S17, the advanced timing control unit generates high-precision, low-jitter timing drive signals according to instructions; S18, using the generated timing drive signal to control multiple independent constant current drive channels.

[0018] The beneficial effects of this application are: unlike the existing technology, the above architecture fundamentally solves the problems of control lag, channel interference, model inaccuracy, and safety rigidity in the existing technology, and realizes high-speed, high-precision, and high-reliability multi-channel light source control. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a schematic diagram of an embodiment of the multi-channel light source controller provided in this application; Figure 2 This is a schematic diagram of the structure of an embodiment of the multi-channel cooperative control unit provided in this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the light source characteristic model unit provided in this application; Figure 4 This is a schematic diagram of the structure of an embodiment of the channel coupling compensation unit provided in this application; Figure 5 This is a schematic diagram of the structure of an embodiment of the intelligent security protection unit provided in this application; Figure 6 This is a schematic diagram of the structure of an embodiment of the advance timing control unit provided in this application; Figure 7 This is a schematic diagram of another embodiment of the multi-channel light source controller provided in this application; Figure 8 This is a schematic diagram of the structure of an embodiment of the complex task execution engine provided in this application; Figure 9 This is a schematic diagram of the structure of an embodiment of the online self-calibration module provided in this application; Figure 10 This is a flowchart illustrating an embodiment of the control method for the light source controller provided in this application; Figure 11 This is a flowchart illustrating another embodiment of the control method for the light source controller provided in this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It is understood that the specific embodiments described herein are only for explaining this application and not for limiting it. Furthermore, it should be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all structures. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0021] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0022] Existing controllers generally employ a microprocessor-based architecture with multiple constant current drives, adjusting the brightness of each channel via PWM or DAC technology. However, independent control of each channel's light source results in poor synchronization, low control precision, and difficulty in implementing complex multi-angle, multi-spectral synchronous triggering lighting schemes, leading to significant limitations in application. This application aims to create a high-speed, high-precision, highly synchronized, highly secure, and adaptive multi-channel light source controller.

[0023] First, this application provides a multi-channel light source controller.

[0024] In some embodiments, see Figure 1 , Figure 1 This is a schematic diagram of an embodiment of the multi-channel light source controller provided in this application. Wherein: The multi-channel light source control includes: a main microcontroller (MCU) 10 and a lead timing control unit 20. The main microcontroller 10 integrates a multi-channel collaborative control unit 11, a light source characteristic model unit 12, a channel coupling compensation unit 13, an intelligent safety protection unit 14, and multiple independent constant current drive channels 30 (this application uses 8 channels as an example for illustration).

[0025] Based on considerations of functional division of labor and cost-effectiveness in practical application scenarios, as well as the flexibility and scalability of the system, the input interface of the light source controller in this application is configured into two types at the hardware level: one is a configuration instruction interface, connected to the main microcontroller 10, used to receive non-real-time or soft real-time upper-level instructions; the other is a hard real-time trigger interface, directly connected to the advanced timing control unit 20, used to receive external synchronization signals requiring nanosecond or microsecond-level time determinism. That is, the core principle of this invention is "routing instructions according to real-time requirements": hard real-time trigger signals with extremely high timing determinism requirements (such as camera exposure pulses) are directly routed to the advanced timing control unit 20 to ensure nanosecond-level synchronization; while all non-real-time instructions requiring complex calculations and decisions (such as brightness settings) are handled by the MCU, thereby achieving a balance between functional complexity and timing accuracy. This design covers two basic operating modes of the controller in different application scenarios, protecting scenarios requiring extreme performance while also covering scenarios requiring conventional control.

[0026] Among them, the multi-channel collaborative control unit 11, the light source characteristic model unit 12, the channel coupling compensation unit 13, and the intelligent safety protection unit 14 constitute an important intelligent collaborative system of this application, which is used to improve the light source controller's high speed, high precision, high synchronization, high safety, and high adaptability in the field of intelligence.

[0027] See Figures 2 to 6 : Figure 2 This is a schematic diagram of the structure of an embodiment of the multi-channel cooperative control unit provided in this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the light source characteristic model unit provided in this application; Figure 4 This is a schematic diagram of the structure of an embodiment of the channel coupling compensation unit provided in this application; Figure 5 This is a schematic diagram of the structure of an embodiment of the intelligent security protection unit provided in this application; Figure 6 This is a schematic diagram of an embodiment of the advance timing control unit provided in this application. The following is in conjunction with... Figures 2 to 6 Each unit will be described in detail: Among them, the multi-channel cooperative control unit 11 is used to temporarily bind multiple channels that need to be synchronized into an inseparable coupled system at the dynamic control level when multiple channels need to be synchronized based on upper-level instructions. In a joint state space composed of all related channels, it uniformly plans an optimal path that strictly coordinates the entire dynamic trajectory from the starting point to the end point, and generates instantaneous target values ​​that are coordinated in time.

[0028] Technical Notes: Traditional control can only ensure that the channels are roughly consistent at the start. However, due to differences in the physical characteristics of each channel (drive circuit response speed, LED load characteristics) and the performance of the feedback loop, their entire dynamic change process (transition trajectory) from the start to the end is mismatched. This asynchrony in the dynamic process can cause serious problems in applications with extremely high consistency requirements. For example, when synthesizing colored light, the inconsistent brightness change rates of the red, green, and blue channels can lead to undesirable intermediate color shifts during the color transformation process.

[0029] This multi-channel cooperative control unit 11 solves the dynamic process mismatch problem caused by parallel independent control in existing technologies through a "holistic" control approach. It virtually binds multiple channels requiring synchronization into a coupled system in a joint state space, and ensures that the output of each channel strictly maintains preset constraints such as brightness ratio and phase at every instant of state change by uniformly planning the optimal cooperative trajectory from the starting point to the end point. This achieves true process-level synchronization.

[0030] Specifically, the multi-channel cooperative control unit 11 is equipped with a synchronous task instruction parser 111, a joint state space and trajectory planner 112, a real-time interpolator 113, and an instruction distribution module 114. Among them: The synchronization task instruction parser 111 is used to parse instructions containing target values, duration, and synchronization mode. Specifically, the synchronization task instruction parser 111 receives JSON-formatted synchronization control instructions from the upper-layer application via an Ethernet interface. The module maintains an instruction parsing state machine, and the parsing process is initiated when the SYNC_CMD instruction header is detected. The typical instruction format is: SYNC_CMD(channel_mask, mode, target_values, duration, profile). First, the channel_mask bitmask (e.g., 0x07) is extracted to determine the set of participating channels. Then, the mode field is read to parse the synchronization mode (e.g., the enumeration value 0x02 corresponding to the chroma preservation synchronization mode). Next, the target brightness values ​​of each channel are loaded from the target_values ​​array (e.g., [255, 128, 64]), and the duration parameter (e.g., 500ms) and profile trajectory type (e.g., parameter 0x03 corresponding to the S-curve) are obtained. All parsing results are encapsulated into a synchronization task descriptor structure and sent to the joint state space and trajectory planner 112 via a message queue. This implementation supports instruction parsing within 100μs, ensuring real-time response performance.

[0031] Subsequently, the joint state space and trajectory planner 112 receives the instructions parsed by the aforementioned synchronization task instruction parser 111. The message is activated to plan a smooth trajectory from the current state to the target state in the joint state space consisting of the channels participating in the synchronization.

[0032] Specifically, the process begins by defining a state vector: an M-dimensional state vector Y(t) = [y1(t), y2(t), ..., yM(t)]^T, where M is the number of channels participating in synchronization, and yi(t) is the instantaneous brightness value of the i-th channel at time t. Next, path planning is performed. Based on the duration and profile parameters in the instructions, the planner calculates a smooth path Y(t) from the current state vector Y_current to the target state vector Y_target in the joint state space, where t ranges from 0 to duration. For example, for "linear interpolation synchronization," this path is: Y(t) = Y_current + (t / duration) * (Y_target - Y_current). Finally, additional constraints are applied based on the mode parameter during path planning. For example, in the "chromaticity synchronization" mode (for RGB color light sources), the planner ensures that the proportional relationship between y_r(t), y_g(t), and y_b(t) on the entire trajectory Y(t) is always consistent with the proportional relationship of the final target value, thereby ensuring that the color purity does not shift during the change process.

[0033] Subsequently, the real-time interpolator 113 is used to sample the trajectory Y(t) to generate the instantaneous target value for each control cycle. The sequence. Specifically, the joint state space and trajectory planner 112 generates a continuous-time ideal path Y(t). The real-time interpolator 113 samples along this ideal path according to the controller's own control cycle (e.g., once every 100 microseconds), generating a series of discrete, instantaneous target values. That is, in each control cycle k, the real-time interpolator 113 calculates the current time t_k and samples the coordinates corresponding to t_k on the continuous trajectory Y(t), thereby obtaining the instantaneous target vector Y_sp(k)=[y1_sp(k),y2_sp(k),...,yM_sp(k)]^T for that control cycle. Finally, this set of instantaneous target values ​​Y_sp(k) obtained through real-time sampling will serve as the temporary control setpoints for all participating synchronization channels within that control cycle.

[0034] Then, the instruction distribution module 114 is used to distribute the instantaneous target value to the underlying controller of each channel. The underlying controller may be the light source characteristic model unit 12.

[0035] Example of implementation process: 1. Task Instruction: The user needs to smoothly change the RGB light source from white [100,100,100] to pure red [255,0,0] within 500 milliseconds via the host computer, and the color should remain unchanged during the process. To this end, the host computer sends a high-level synchronization instruction to the light source controller: SYNC_CMD(mask=0b111,mode=chroma preservation,target=[255,0,0],duration=500ms,profile=linear).

[0036] 2. Unit activation: After receiving the instruction, the multi-channel collaborative control unit 11 is activated, and the synchronous task instruction parser 111 performs instruction parsing.

[0037] 3. Trajectory Planning: The joint state space and trajectory planner 1122 then perform collaborative path planning in the joint state space (i.e., the three-dimensional color space) composed of the R, G, and B channels. Since the instruction mode is "chroma preservation", the planner does not simply calculate a straight line from the starting point to the ending point, but plans a dynamic curve that ensures that the decay rate of the G and B channels and the enhancement rate of the R channel are precisely matched in colorimetry, thereby ensuring that the hue and saturation of the color do not shift during the entire 500-millisecond transition period.

[0038] 4. Real-time Interpolation and Closed-Loop Execution: Assuming the system control cycle is 1 millisecond, in the next 500 cycles: Cycle 1 (t=1ms): The real-time interpolator 113 performs its first sampling from the planned cooperative trajectory, obtaining the instantaneous target vector [100.31, 99.8, 99.8]. This set of fine-tuned target values ​​is simultaneously distributed to the underlying controllers of the three channels. Each channel calculates the precise drive current and executes accordingly, realizing the first step of cooperative change. Cycle 250 (t=250ms): The interpolator samples near the midpoint of the path, obtaining the instantaneous target [177.5, 50, 50]. At this time, the three channels have cooperatively changed to the intermediate state. Cycle 500 (t=500ms): The interpolator performs the final sampling, obtaining the final target [255, 0, 0]. The system precisely reaches the specified pure red state.

[0039] 5. Task termination and authority transfer: After the synchronization task is completed, the multi-channel collaborative control unit 11 automatically returns the control of each channel to the regular control logic, and the three channels are stably maintained in the final state of [255,0,0].

[0040] Among them, the light source characteristic model unit 12 is used to establish a dynamic model describing the photoelectric and thermal characteristics of the light source load for each channel, and to solve the above instantaneous target value output in reverse according to the model to realize closed-loop control from the electrical domain to the optical domain.

[0041] Technical Notes: The light source characteristic model unit 12 elevates the control objective from traditional stable current to direct and precise control of optical output by establishing a dynamic white-box model that includes internal state variables such as junction temperature and aging degree. This model, based on forward prediction, can simulate the actual luminous flux and color temperature of the LED under a given drive current in real time. Through inverse solving, it converts the user-defined optical objective into precise drive current commands, thereby achieving closed-loop optical control unaffected by changes in operating conditions throughout the entire operating range and lifespan.

[0042] Specifically, the light source characteristic model unit 12 is configured with a model parameter storage library 121, a real-time state observer 122, and a front... To predictor 123 and reverse control command generator 124, wherein: Model parameter repository 121 is used to pre-store a set of model parameters obtained through offline calibration and knowledge distillation for each supported LED model. Specifically, a separate parameter storage area is reserved for each supported LED model in the flash memory or electrically erasable programmable read-only memory (EEPROM) inside the light source controller. These parameters are obtained through a high-precision offline calibration process: in an integrating sphere measurement system, continuous scanning excitation from zero to rated current is applied to different LED models, and their luminous flux, color temperature, and junction temperature data are collected simultaneously. The core parameters of the state-space model are extracted through a system identification algorithm, including the state transition matrix A, input matrix B, output matrix C, and direct action matrix D describing the dynamic characteristics of the system; at the same time, a set of parameters for establishing nonlinear mapping relationships is established, including a current-luminous efficacy lookup table stored using piecewise linearization, and third-order polynomial coefficients characterizing the junction temperature-spectral drift relationship; thermal characteristic parameters are also included, such as the thermal resistance from the chip to the substrate and the system thermal capacity; finally, aging model parameters are stored, including exponential function coefficients characterizing light decay characteristics and lifetime decay curve parameters. All parameters are stored in the form of a structure, which is loaded into the main microcontroller's memory when the system starts for real-time calculation and access.

[0043] Example of offline calibration process: First, a teacher model is constructed. The LED to be calibrated is placed on a temperature-controlled platform equipped with a spectrometer and integrating sphere. A driving current of 0-1500mA is applied using a Keithley 2600 series high-precision source meter. Simultaneously, complete spectral power distribution, luminous flux, and forward voltage data are recorded within a stable junction temperature range of 25℃ to 85℃. The light decay characteristic curve is obtained through a 1000-hour accelerated aging experiment. Based on this massive amount of experimental data, the recursive least squares method with a forgetting factor is used to identify the parameters of the student model: discrete state-space model x(k+1)=Ax(k)+Bu(k); y(k)=Cx(k)+Du(k). The state variable x mainly includes the junction temperature and its rate of change, the input u is the driving current, and the output y is the luminous flux and color temperature coordinates. The final extracted core parameters include: state transition matrix A (2×2), input matrix B (2×1), output matrix C (2×2), direct action matrix D (2×1), current-optical efficiency lookup table (256 points) characterizing nonlinear relationships, junction temperature-wavelength drift polynomial coefficients (third order), thermal resistance parameter (θJA=8℃ / W) and aging decay coefficient (β=-0.0012 / kh), which together constitute a complete set of model parameters. After verification, these parameters are solidified and stored in a specified sector of Flash.

[0044] The real-time state observer 122 is used to maintain a current state vector for each channel in real time. It iteratively updates the internal state by acquiring the actual output current of the driving circuit and the temperature of the LED substrate in real time, combined with internal thermal and aging models. Specifically, an accurate state vector x(k) is maintained for each channel in the following way: During the input acquisition phase, the real-time state observer 122 acquires the output current I_actual of the driving circuit in real time at a sampling rate of 100kHz (from a 50mΩ sampling resistor and an INA240 current sense amplifier), and simultaneously reads the temperature T_board of the NTC thermistor attached to the LED substrate via the SPI interface (accuracy ±0.5℃). During the state update phase, a state estimation is triggered once every control cycle (e.g., 100μs): based on the state vector x(k-1) of the previous moment and the currently acquired driving current I_actual, a matrix operation of the state transition equation x(k) = A·x(k-1) + B·u(k-1) is performed, where u(k-1) is the driving command of the previous cycle. For junction temperature estimation, the accuracy is significantly improved by combining real-time acquired T_board data with the internal thermal model. Simultaneously, an internal timer accumulates the channel illumination time, and this time parameter is input into the aging model to calculate the current luminous efficacy attenuation coefficient in real time.

[0045] The forward predictor 123 receives a hypothetical drive current value and calculates the predicted optical parameters that the LED will produce under that hypothetical current. Specifically, after receiving a hypothetical drive current value I_hypothetical, the forward predictor 123 immediately initiates the prediction calculation process. First, based on the current state vector x(k) and the input hypothetical current value, it performs matrix operations on the output equation y_pred(k) = C·x(k) + D·I_hypothetical(k) to obtain the linear basic optical parameters. Subsequently, it calls the nonlinear mapping function stored in the model parameter repository 121 for precise correction: nonlinear calibration of the basic luminous flux is performed by querying a 256-point current-luminous efficacy lookup table, while temperature drift compensation of the color temperature parameters is performed using third-order polynomial coefficients. Finally, the predicted luminous flux L_pred (in lumens) and predicted color temperature CCT_pred (in K) calculated by the complete model are output. These parameters accurately reflect the true optical characteristics of the LED under the given hypothetical current and current thermal conditions.

[0046] The reverse control instruction generator 124 iteratively calls the forward predictor to solve for the drive current value that matches the predicted optical output with the user-defined target. For example, when the user sets the optical target (e.g., "the output luminous flux of the second channel is 1200 lumens") via the Ethernet interface, the reverse control instruction generator 124 initiates the reverse solution process. First, an improved binary search method is used for iterative optimization: with 50% of the rated drive current as the initial guess value I_guess, the forward predictor module is called to calculate the corresponding luminous flux L_pred. When |L_pred-1200|>5 lumens, the current guess value is adjusted in 10mA steps according to the error sign, and the iterative calculation is repeated. The entire iterative process is accelerated using the FPU unit of the ARM Cortex-M7 core, achieving convergence within 50 microseconds. When |L_pred-1200|≤5 lumens is satisfied, the final current value I_target is immediately written to the DAC register of the constant current driver chip via the SPI interface, achieving precise conversion from the optical target to the drive instruction.

[0047] Example of implementation process: The user sets the brightness of channel 2 to 80%. The controller internally calibrates that this LED model, under standard conditions (25 degrees Celsius, brand new), corresponds to a luminous flux of 1000 lumens at 80% brightness. 1. Initial Stage: The system first retrieves the complete parameter set of the LED model (including state space matrices A / B / C / D, nonlinear lookup tables, thermal parameters, and aging coefficients) from the model parameter repository 121. The reverse control command generator 124 performs iterative calculations based on these parameters and the initial cold-state conditions. Through multiple calls to the forward predictor 123 for simulation, the forward predictor 123 calculates, based on matrices C and D and the current-luminous efficacy lookup table in the model parameter repository 121, that only a 700mA driving current is needed to achieve the target of 1000 lumens under low-temperature conditions. This current command is then output to the driving circuit.

[0048] 2. Status Monitoring Phase: The real-time status observer 122 operates continuously, using the thermal model parameters (thermal resistance, heat capacity) in the model parameter storage library 121, combined with the collected temperature sensor data, to dynamically update the system state vector. When the LED junction temperature is detected to have risen to 60 degrees Celsius, the new state parameters are immediately transmitted to the entire system.

[0049] 3. Dynamic Compensation Phase: In the next control cycle, the reverse control command generator 124 starts calculation again with a target of 1000 lumens. At this time, when the forward predictor 123 is called, based on the parameters provided by the same model parameter repository and combined with the updated high temperature conditions, it is predicted that the original 700mA current can only produce 950 lumens output at the current temperature.

[0050] 4. Closed-loop adjustment phase: After the reverse control command generator 124 detects a prediction deviation of 50 lumens, it automatically restarts the iterative optimization process. Through multiple calls to the forward predictor 123 for simulation calculations, all prediction calculations are based on authoritative parameters in the model parameter repository. Finally, it is determined that the drive current needs to be increased to 725mA to compensate for thermal light decay. The controller then smoothly adjusts the output current to the new set value.

[0051] Therefore, through the coordinated operation of each module, the system completes real-time compensation for temperature drift without the user's awareness. External users always observe a stable 80% brightness output, while the light source controller has achieved a fundamental shift from "current control" to "luminance control" through a dynamic state-space model, ensuring output accuracy and stability across the entire operating range and throughout the LED's lifespan.

[0052] The channel coupling compensation unit 13 is used to predict the above-mentioned driving commands based on a pre-established multi-input multi-output transfer function model that describes the physical coupling relationship between channels, and generate a feedforward compensation signal to suppress mutual interference between channels; the parameters of the transfer function model are determined by measuring the mutual interference characteristics between channels during the design or production calibration stage of the controller.

[0053] Technical Notes: Physically, all channels share the same power supply system and circuit board environment. When the current in any channel changes drastically, it will cause a bus voltage sag and electromagnetic crosstalk through the power supply internal resistance and line impedance. Unlike the passive compensation of existing technologies that rely on hysteresis feedback, this application introduces the concept of feedforward control, treating the interference as a predictable deterministic disturbance: First, based on a preset inter-channel coupling function model, the degree of disturbance to the power supply environment of other channels is predicted in real time according to the drive command to be executed; then, a set of compensation signals of equal magnitude and opposite direction are immediately generated and fed forward to the control loop of the disturbed channel and superimposed on the original command, thereby canceling the influence of the interference at the source before it occurs, ensuring that the photometric output of each channel can remain stable under high-speed transient conditions.

[0054] Specifically, the channel coupling compensation unit 13 is configured with a multi-input multi-output coupling interference transfer function model 131, a drive instruction preprocessor 132, a real-time disturbance predictor 133, and a compensation signal injection module 134. Wherein: The coupling interference transfer function model 131, obtained through system identification during the controller design or production calibration phase, describes the voltage disturbance caused by a change in current in any channel to the remaining channels. For example, an N×N coupling interference transfer function model H(s) is stored in the controller firmware or non-volatile memory. For an eight-channel controller, this model is 8×8 in dimension. Each element Hij(s) in the model (where i≠j) is a transfer function that precisely describes the equivalent voltage disturbance ΔVi caused at the drive node of the i-th channel when the drive current Ij of the j-th channel changes.

[0055] The specific implementation method is as follows: Keep all channels closed, apply a known current step or sweep frequency signal dIj / dt to channel j, and simultaneously measure the drive node voltage disturbance waveform ΔVi(t) of all other channels i using a high-precision oscilloscope or data acquisition card. By analyzing these input-output data (using the least squares method or frequency domain analysis method), the coefficients of the transfer function Hij(s) are fitted. In the specific implementation, this function is usually simplified to a first-order or second-order low-pass or band-pass filter model. This simplified form facilitates discretization and real-time calculation in the microcontroller. In actual operation, this model is executed in real time in the form of discretized difference equations in the microcontroller, which can predict the coupling interference effect between channels according to the changes in the drive commands of each channel.

[0056] The driver instruction preprocessor 132 is used to calculate the expected current change rate of each channel's driver instruction. Specifically, this preprocessor 32 is integrated into the interrupt service routine of the main microcontroller and completes preprocessing before all driver instructions are issued to the hardware. The specific execution flow is as follows: The driver instruction preprocessor 132 receives the target driver instruction array Target_I[0...7] from the upper-layer application logic in real time, and simultaneously obtains the current actual current value array Current_I[0...7] of each channel through ADC sampling. At the beginning of each fixed control cycle T_control (typically 100 microseconds), the module performs calculations for each channel j: first, it calculates the difference between the target value and the current value (Target_I[j] - Current_I[j]), and then divides this difference by the control cycle T_control to obtain the expected current change rate dI_pred[j] / dt of each channel in the next control cycle. These calculated change rate values ​​are stored in a dedicated array buffer for subsequent disturbance prediction modules to call in real time.

[0057] The real-time disturbance predictor 133 is used to predict the equivalent voltage disturbance that all channels will experience based on the aforementioned transfer function model and the expected rate of change of current. Specifically, the real-time disturbance predictor 133 serves as the computational core of the channel coupling compensation unit 13, and performs the following operation in each control cycle: For each channel i (from 0 to 7), the total equivalent voltage disturbance ΔV_pred[i] that the channel will experience is first calculated. This calculation is achieved by taking the expected rate of change of current dI_pred[j] / dt of all other channels j (j≠i) as input and substituting it into the i-th row of the coupling disturbance transfer function model H(s) stored in the controller for convolution operation. The specific calculation formula is: ΔV_predi = Σ(j≠i)[Hij(s)*(dI_pred[j] / dt)]. In the discretized implementation of the main microcontroller, this convolution operation in the continuous domain is simplified to an iterative calculation of a first-order difference equation. Each transfer function Hij(s) is discretized into a digital filter form, and the predicted voltage disturbance value can be obtained through iterative calculation. After obtaining the predicted voltage disturbance ΔV_pred[i], the feedforward compensation value Comp[i] to be applied to the i-th channel control loop is calculated based on the characteristic parameter K_v_i of the constant current drive circuit (i.e., the influence coefficient of the input voltage change on the output current): Comp[i] = -K_v_i * ΔV_pred[i]. The compensation value is negative to produce an effect opposite to the predicted disturbance, thus canceling it out when it occurs. The compensation values ​​of all eight channels together form a compensation vector, which is output to the subsequent instruction synthesis module for further processing.

[0058] The compensation signal injection module 134 is used to calculate the feedforward compensation value to be applied to each channel control loop based on the predicted voltage disturbance and the characteristics of the constant current drive circuit, and synthesize the compensation value with the original drive command to generate the anti-interference final drive command. Specifically, in each control cycle, the module receives the original drive command array Target_I[0...7] from the upper-level application logic and the feedforward compensation vector Comp[0...7] from the real-time disturbance predictor 133, and performs vector synthesis operation through a parallel adder array to generate the preliminary drive command: Final_I[i]=Target_I[i]+Comp[i]. At the same time, the independent closed-loop PID feedback controller of each channel continues to run, and outputs the feedback correction amount Feedback[i] based on current sampling in real time. Finally, the module superimposes the feedforward compensation and the feedback correction to form the complete drive signal Output_I[i]=Final_I[i]+Feedback[i]. This composite control structure actively eliminates inter-channel coupling interference through feedforward compensation and effectively compensates for model prediction errors and steady-state errors caused by uncoupled factors through feedback control, thereby achieving high-precision multi-channel collaborative control.

[0059] Example of implementation process: Assuming the controller is in a stable operating state, all eight channels are constantly lit at 50% brightness. At this time, the system receives a control command requiring channel 1 to instantaneously turn on to 100% brightness in the next control cycle, while keeping the brightness of the other channels unchanged.

[0060] I. Work status of technologies without this application: The drive circuit of channel 1 will momentarily draw a large current from the power bus. Due to the internal resistance of the power supply, this causes the bus current to... The voltage drops briefly. This voltage drop is simultaneously transmitted to the drive circuits of the other 7 channels, causing a decrease in their actual driving capability and a reduction in output current. This results in a noticeable momentary dimming of the light source in these channels, which then slowly recovers to the set brightness under the action of the feedback loops of each channel.

[0061] II. Workflow using the technology described in this application: 1. Instruction preprocessing: After the control instruction "Channel 1 rises to 100%" is intercepted by the preprocessor, the expected rate of change of current of Channel 1, dI_pred[1] / dt, is calculated to be a large positive value, while the dI_pred[j] / dt (j≠1) of other channels are all zero.

[0062] 2. Disturbance Prediction: The disturbance predictor immediately starts its operation. It calls the 8x8 coupled disturbance transfer function model H(s) pre-stored in the system. Specifically, for the second channel, the convolution of element H21(s) with dI_pred[1] / dt is calculated, i.e., ΔV_pred[2]=H21(s)*(dI_pred[1] / dt), thereby accurately predicting the equivalent voltage disturbance that the action of the first channel will cause on the second channel. Similarly, H31(s) to H81(s) in the model are called synchronously to calculate the voltage disturbances that the third to eighth channels will face.

[0063] 3. Compensation generation: Based on the voltage disturbance data (ΔV_pred[2] to ΔV_pred[8]) predicted by the above model, the module combines the voltage-current coefficient K_v_i of each driver to generate a set of negative feedforward compensation values ​​Comp[2...7] for channels 2 to 8.

[0064] 4. Instruction Synthesis and Execution: The final driver instructions sent to the lower layer are as follows: the target value of channel 1 is set to 100%; the target values ​​of channels 2 to 8 are based on the original 50% with their respective compensation values ​​Comp added.

[0065] 5. Control Effect: When a large current is drawn from channel 1, causing a drop in bus voltage, channels 2 through 8, having received compensation commands calculated in advance based on model predictions, have their drivers operating at even higher power, perfectly offsetting the decrease in driving capability caused by the voltage drop. External observation shows that the brightness of the light source in channels 2 through 8 remains almost unchanged throughout the entire dynamic process, achieving true dynamic isolation between channels.

[0066] The intelligent safety protection unit 14 is used to predict the execution consequences of the compensated driving instructions based on a multi-dimensional safety boundary function, and to minimize the correction of the instructions when they may exceed the safe area, so as to generate safe final driving instructions.

[0067] Technical Notes: The intelligent flexible protection system based on the safety boundary function theory represents a revolution in the traditional "static red line" safety mechanism. Traditional protection relies on fixed thresholds, which suffers from three major drawbacks: conservatism, rigid interruption, and single-dimensional judgment. This application constructs a "multi-dimensional dynamic safety domain" defined by all key system state variables (such as current, voltage, junction temperature, etc.). It utilizes a safety boundary function, whose value is positive within the safety domain, zero at the boundary, and negative in the danger domain. By predicting the driving command in advance: if the predicted state is within the safety domain, it is allowed; if the prediction indicates an out-of-bounds state, an online optimization process is initiated. The goal is to calculate the correction scheme with minimal changes to the original command, guiding the system state back within the safety boundary.

[0068] Specifically, the intelligent safety protection unit 14 is equipped with a multi-dimensional safety boundary function definition module 141, a system state prediction module 142, a hazard assessment module 143, and an online optimizer 144. Among them: The multidimensional safety boundary function definition module 141 is used to define a safe region composed of multiple state variables, including current, voltage, temperature, and power change rate. Specifically, a set of boundary functions h_j(X) for defining the safe operating region is first stored in the non-volatile memory of the light source controller, where X is a vector representing the entire controller state. The state vector X is constructed as follows: X=[I_1,...,I_8,V_bus,T_1,...,T_8,dI_1 / dt,...] contains key state variables such as the real-time current of all 8 channels, the shared bus voltage, the real-time junction temperature of each channel's LED (estimated by the model), and the instantaneous rate of change of each channel's current.

[0069] The specific forms of the boundary functions include: 1. The overcurrent boundary function h_ocp_i(X) = I_max(T_i) - I_i, where I_max is no longer a constant, but a function related to the junction temperature T_i, reflecting the high-temperature derating usage. 2. Over-temperature boundary function h_otp_i(X) = T_j_max - T_i (T_j_max is the highest allowable junction temperature of the LED); 3. The total power boundary function h_opp(X) = P_total_max - Σ(I_iV_i) (limits the total power consumption of the entire controller); 4. The dynamic safety boundary function h_slew_rate(X) = S_max - |dI_i / dt| (prevents excessively rapid current changes from impacting the power supply or generating excessive electromagnetic interference). Ultimately, the entire safe region S is defined as the state space region where all boundary function values ​​are greater than or equal to zero, i.e., S = {X|h_j(X) ≥ 0, for allj}.

[0070] The system state prediction module 142, tightly coupled with the aforementioned light source characteristic model unit 12, predicts the future system state after executing the compensated drive instruction. Specifically, in each control cycle: first, the system generates a drive instruction vector U_comp that includes inter-channel coupling feedforward compensation, which is about to be sent to the hardware. Upon receiving U_comp, this module immediately performs the core operation, namely forward simulation: taking the current system state X(k) and the compensated drive instruction U_comp as input, it calls the dynamic state space model to perform a single-step forward simulation calculation, thereby predicting the system state X_pred(k+1) that will be reached at the end of the next control cycle when the compensated instruction is executed, and making the final safety decision accordingly.

[0071] Subsequently, after receiving the predicted future state X_pred(k+1), the risk assessment module 143 substitutes the vector into all the safety boundary functions h_j(X) for evaluation, that is: h_j_pred=h_j(X_pred(k+1)). If h_j_pred in all h_j(X) is greater than a small positive safety margin, it is evaluated as safe. Conversely, when it is predicted that the future system state may exceed the safe area, it is evaluated as having potential risk, and the online optimizer 144 is triggered to start an optimization process to calculate a correction amount that minimizes the change to the original instructions.

[0072] Among them, the online optimizer 144, when there is a risk, solves a solution that aims to minimize the instruction correction amount and satisfy the safety requirements. This is an optimization problem with full boundary constraints, outputting a corrected safety instruction. Specifically, the online optimizer 144 will initiate a mini online optimization calculation, the core objective of which is to find an optimal correction vector U_corr such that the final executed instruction U_final = U_comp + U_corr, while satisfying all safety constraints, minimizes the difference between it and the compensated instruction U_comp to be executed. This optimization problem is constructed as a standard quadratic programming problem, with the optimization objective being to minimize the correction amount min||U_corr||², to ensure that the correction action minimizes the perturbation to the original instruction. The constraints are based on the first-order Taylor expansion of the boundary function, linearizing the fundamental constraint h_j(X(k+1))≥0, which requires the system state to be within the safety region after executing the final instruction, into a linear inequality with respect to the correction vector U_corr: h_j(X(k))+L_gh_j(X(k))×(U_comp+U_corr)+L_fh_j(X(k))≥-α(h_j(X(k))), where L_gh_j and L_fh_j are the Lie derivatives (i.e. rates of change) of the boundary function h_j with respect to the control input and the natural dynamics of the system, and α is a K-like function that ensures that the state converges to the safe region. Solving this problem ensures the safety of the final driving instructions.

[0073] Then, this safe final instruction is passed to the advanced timing control unit 20.

[0074] In some embodiments, the lead timing control unit 20 is configured to receive external drive commands through two paths: one is to receive modified drive commands from the main microcontroller 10 (soft real-time trigger interface) and generate high-precision, low-jitter timing drive signals; the other is to directly receive external synchronization signals from the hard real-time trigger interface and, based on the external synchronization signals, independently of the main microcontroller 10, convert the external synchronization signals into high-precision, low-jitter timing drive signals. For drive commands / external synchronization signals received from different interfaces, the lead timing control unit will generate high-precision, low-jitter timing drive signals according to the process.

[0075] Technical Notes: The advanced timing control unit 20 is preferably implemented by a field-programmable gate array (FPGA) or complex programmable logic device (CPLD). Its physical location in the circuit is directly connected to both the external trigger signal input interface and the main microcontroller 10 to efficiently process instructions / signals from different interfaces. Its output is directly or indirectly connected to the underlying driving circuit of each light source channel (e.g., pulse width modulation signal generator, trigger pin of digital-to-analog converter, or gate driver of power switch).

[0076] The advanced timing control unit 20 is configured with the following functional logic: an input signal conditioning and arbitration module 21, a timing instruction decoding and parameter register group 22, an atomic timing operation library 23, and a high-precision timing sequence generator 24. Among them: The input signal conditioning and arbitration module 21 has its input end directly connected to the hard real-time trigger interface of the light source controller. It is used to electrically isolate the external trigger signal through an optocoupler, and to filter out high-frequency noise and mechanical jitter of the input signal through a hardware-implemented digital low-pass filter or debouncing logic. It also uses a high-frequency internal clock to capture the high-precision timestamp of the edge of each valid trigger signal. First, an optocoupler is used to achieve electrical isolation of the input signal, and a 16th-order digital low-pass filter based on a shift register and debouncing logic are used to effectively filter out high-frequency noise and mechanical jitter, ensuring a clean and valid trigger edge. Second, a 32-bit counter driven by a 200MHz high-frequency internal clock is used to form a high-precision timestamp capture unit, which accurately timestamps the transition edge of each valid trigger signal, achieving a capture accuracy of 5 nanoseconds. Third, a dedicated event arbitration logic is designed. When multiple external trigger signals arrive simultaneously within one clock cycle (5ns), this logic performs real-time arbitration according to a preset fixed priority rule to determine a unique event trigger sequence. The arbitrated event and its precise timestamp are stored in a 16-depth First-In-First-Out (FIFO) buffer, providing a stable, orderly trigger instruction with precise time information for subsequent timing processing modules.

[0077] The timing instruction decoding and parameter register group 22 is used to receive and store timing task configurations from the main microcontroller 10. This timing instruction decoding and parameter register group 22 establishes a connection with the main microcontroller 10 through a standard bus interface of a 32-bit SPI serial peripheral. The main microcontroller 10 writes advanced timing control task configuration information into the parameter register group inside the timing instruction decoding and parameter register group 22 through this bus. For example, an 8-bit wide mode control register is set to configure the overall operating mode of the timing instruction decoding and parameter register group 22, including external trigger mode, internal software trigger mode, and sequence generation mode; a set of channel parameter registers is independently allocated for each light source channel to store timing parameters such as pulse width, trigger delay, strobe frequency, and duty cycle; a sequence definition register area is also provided to store the complete definition of complex timing sequences. This area sequentially stores a series of atomic timing opcodes and their corresponding operation parameters, forming an instruction sequence that can be directly parsed and executed by hardware.

[0078] The Atomic Timing Operation Library 23 solidifies the fundamental, indivisible atomic timing operation logic. Implemented using a hardware description language and embedded in the FPGA logic resources, it constitutes a complete hardware timing engine. The Atomic Timing Operation Library implements three types of basic timing functions in the form of parameterized modules: for example, a single pulse generation module receives the channel address and pulse width parameters to directly control the output of a single pulse from a specified channel; a pulse train generation module generates a continuous pulse sequence based on frequency and quantity parameters; and a multi-channel synchronization trigger module achieves nanosecond-level synchronization across multiple channels through a unified enable signal.

[0079] The high-precision timing sequence generator 24, based on the atomic timing operation library and configuration parameters, autonomously and accurately generates the final underlying drive control waveform using a high-frequency internal clock and multiple high-precision counters. Specifically, the high-precision timing sequence generator 24 adopts a three-stage state machine architecture. After detecting a valid trigger event, the state machine automatically calls the corresponding functional module in the atomic timing operation library 23 according to the opcode and parameters provided by the timing instruction decoder and parameter register group 22 or the trigger instruction provided by the input signal conditioning and arbitration module 21. At the same time, it enables a network of multiple 32-bit / 64-bit high-precision counters based on a 200MHz clock for precise timing, autonomously completing the entire process from instruction parsing to waveform generation, and finally outputting an underlying drive control waveform with jitter of less than 100ps.

[0080] Finally, an output driver interface module is used to output the logic control signals generated by the timing sequence generator to the corresponding physical pins on the controller, so as to directly drive the underlying circuits of each channel.

[0081] Example of implementation process: This implementation uses high-speed triggered strobe lighting as an application scenario to demonstrate the complete workflow of two paths of the advanced timing control unit: A: Parameter pre-configuration and command decoding (soft real-time triggering interface approach) The host computer sends a configuration command to the main microcontroller 10 via Ethernet, requesting that it be set to externally triggered strobe mode, and specifying when... When the TRIG_IN signal arrives, channels 1 and 3 must synchronously generate a bright pulse with a width of 10.00 microseconds after a 2.00 microsecond delay. After parsing the instruction, the main microcontroller 10 writes parameters such as the mode code, channel mask (0x05), delay time (400 clock cycles), and pulse width (2000 clock cycles) into the registers of the timing instruction decoding and parameter register group 22 via the SPI bus. This step completes the conversion and storage from high-level instructions to hardware-executable parameters. After configuration is complete, the main microcontroller 10 exits the real-time control loop. Subsequently, the atomic timing operation library 23 works in conjunction with the high-precision timing sequence generator 24: the timing sequence generator 24 immediately initiates a hard real-time response process, executing specific tasks by calling the standard atomic operation modules in the atomic timing operation library 23: first, it enables a delay counter to perform precise timing for 400 clock cycles (corresponding to 2.00 microseconds); after the delay, it calls a multi-channel synchronous trigger atomic operation to synchronously set the DRV_OUT_1 and DRV_OUT_3 output pins; simultaneously, it calls a pulse generation atomic operation to start a pulse width counter to perform precise timing for 2000 clock cycles (corresponding to 10.00 microseconds); after the timing is completed, it calls another atomic operation to synchronously clear the two output pins. This execution mechanism based on the atomic operation library ensures the standardization and extremely high reliability of timing operations. Throughout the process, all timing control is completely autonomously completed by the hardware state machine, and the main microcontroller can process non-real-time tasks such as communication and monitoring in parallel.

[0082] B: Event Triggering and Signal Processing (Hard Real-Time Triggering Interface Approach) When the external trigger signal TRIG_IN arrives, it directly enters the input signal conditioning and arbitration module 21. This module 21 processes the signal... The signal undergoes opto-isolation, digital filtering, and debouncing to generate a clean INTERNAL_TRIG event. A 5ns timestamp is then appended before being sent to the high-precision timing sequence generator 24. Subsequently, the atomic timing operation library 23 works in conjunction with the high-precision timing sequence generator 24: the timing sequence generator 24 immediately initiates a hard real-time response process, executing specific tasks by calling standard atomic operation modules in the atomic timing operation library 23. First, a delay counter is activated for precise timing over 400 clock cycles (corresponding to 2.00 microseconds). After the delay, a multi-channel synchronous trigger atomic operation is called to synchronously set the DRV_OUT_1 and DRV_OUT_3 output pins. Simultaneously, a pulse generation atomic operation is called to start a pulse width counter for precise timing over 2000 clock cycles (corresponding to 10.00 microseconds). After the timing is completed, the atomic operation is called again to synchronously clear the two output pins. This execution mechanism based on the atomic operation library ensures the standardization and extremely high reliability of timing operations. Throughout the entire process, all timing control is autonomously completed by the hardware state machine, allowing the main microcontroller to handle non-real-time tasks such as communication and monitoring in parallel.

[0083] See Figures 7 to 9 , Figure 7 This is a schematic diagram of another embodiment of the multi-channel light source controller provided in this application. Figure 8 This is a schematic diagram of the structure of an embodiment of the complex task execution engine provided in this application. Figure 9 This is a schematic diagram of an embodiment of the online self-calibration module provided in this application.

[0084] In some embodiments, the main microcontroller 10 further includes a complex task execution engine 15, used to receive and parse non-real-time or soft real-time upper-layer macro instructions, and autonomously decompose them into execution strategies based on an internally stored task knowledge base; the execution strategies include: For tasks requiring multi-channel dynamic coordination, the multi-channel coordination control unit 11 is invoked according to the generated coordination task instructions. The multi-channel coordination control unit 11 generates the coordination trajectory and instantaneous target. Then, through the joint cooperation of the light source characteristic model unit 12, the channel coupling compensation unit 13 and the intelligent safety protection unit 14, the final drive instruction is output to the advanced timing control unit 20. For basic, independent operations, the advance timing control unit 20 is directly scheduled based on the sequence of atomic operations it decomposes, so as to execute the entire lighting sequence autonomously with high efficiency and precision.

[0085] Specifically, the complex task execution engine 15 includes: a task knowledge base 151, an instruction parser and decomposer 152, an atomic illumination operation library 153, and a real-time scheduling executor 154. Among them: Task Knowledge Base 151 stores the definitions of multiple complex lighting tasks and their scripts for decomposing them into atomic operations. Each task definition includes: a task name (macro instruction), a parameter list (such as exposure events, channel lists, etc., as well as their default values ​​and value ranges), and decomposition rules (execution script). This script is a dedicated, interpreted microcode consisting of a series of calls to atomic operations.

[0086] Instruction parser and decomposer 152 is used to receive macro instructions, retrieve corresponding tasks from the knowledge base, and generate an atomic operation team. The instruction parser and decomposer 152 captures macro instruction strings conforming to a predetermined format (such as "EXECUTEPhaseShift_4_Steps(exposure_time_ms=30,channel_A=3,channel_B=4)") by listening to communication interfaces such as Ethernet. After activation, it first searches for matching task templates in the internal task knowledge base, and then binds the actual parameters in the instruction (such as 30) to the corresponding variables (such as $exposure_time_ms) in the template script. Finally, it generates an execution queue consisting of a series of atomic operation instructions with specific parameters bound to them, and submits it to the real-time scheduler executor 154, thereby completing the conversion from high-level macro instructions to executable low-level action sequences.

[0087] The atomic illumination operation library 153 contains optimized low-level control functions for the real-time scheduling actuator to call; these functions are implemented in C functions or assembly code within the firmware code of the optical controller and are also called by the real-time scheduling actuator 154. For example: `voidop_set_brightness(uint8_tchannel, uint16_tvalue)`: Sets the brightness value for the specified channel. `voidop_delay_us(uint32_tduration)`: Implements precise delays at the microsecond level.

[0088] voidop_sync_trig_out(uint8_tsignal_id): Generates a pulse on a synchronization output pin to trigger camera exposure.

[0089] voidop_start_strobe(uint8_tchannel,uint32_tfreq_hz,uint32_tcount): Starts hardware strobe on the specified channel.

[0090] The real-time scheduling executor 154 is used to sequentially retrieve and execute instructions in the atomic operation queue. Specifically, it runs as a background task or interrupt service routine. Its implementation is based on a queue-driven mechanism: this module extracts atomic operation instructions from the execution queue in a first-in-first-out manner, maps the instruction opcode (such as "SET_BRIGHTNESS") to the corresponding function address in the atomic operation library by querying a predefined function pointer table; during execution, a dual-mode execution strategy is adopted, using blocking waiting for delayed operations to ensure timing accuracy, and using non-blocking startup for hardware function operations to achieve concurrent processing; by setting it as the highest real-time task priority, interference from non-real-time tasks such as network communication is effectively isolated, thereby ensuring the microsecond-level timing accuracy of complex lighting sequences.

[0091] In some embodiments, the light source controller further includes an online self-calibration module 16 for incorporating feedback from an external vision system. The signal is compared with the predicted value of the internal light source characteristic model unit, and the model parameters are automatically updated through optimization algorithms to compensate for the performance drift of the light source load caused by aging or environmental changes.

[0092] Technical Notes: The online self-calibration module 16 first actively generates a small driving signal disturbance and injects it into the system, while simultaneously acquiring external visual feedback (such as image grayscale values). Then, it compares the actual feedback changes with the predicted values ​​of the internal dynamic state-space model to calculate the deviation. Finally, using the deviation as the loss function, it optimizes and corrects the model parameters of the light source characteristic model unit 12 online through a gradient descent algorithm, enabling the controller to have lifelong learning capabilities. Wherein: The online self-calibration module 16 is configured with: an external feedback interface 161, an exploratory signal generator 162, a predictor based on an internal model 163, and a model optimizer 164. Among them: External feedback interface 161 is used to receive image quality feedback signals from the vision system. Specifically, external feedback interface 161 receives external visual feedback data by defining a standard communication protocol (such as JSON data packets based on Ethernet / serial port, including channel number, feedback type and value), and the built-in data processor performs real-time preprocessing (including digital filtering and normalization) on the raw feedback values. At the same time, it accurately associates and stores the feedback with the internal driving command that triggered the feedback by using timestamps or event identifiers, thereby establishing a causal data pair of driving and feedback for model optimization.

[0093] An exploratory signal generator 162 is used to apply a small drive perturbation near the stable operating point. Specifically, after the self-calibration mode is triggered, the user-defined stable reference drive current I_base is first locked as the operating point; then, a set of small orthogonal or random perturbation signals ΔI_k are generated around this operating point, which is achieved by superimposing a multi-frequency sine wave signal with a small amplitude or an alternating positive and negative step signal on I_base; finally, the composite drive command I_base+ΔI_k is injected into the underlying drive circuit in a very short time, and the command issuance timestamp and perturbation parameters are recorded simultaneously to establish drive-response correlation data.

[0094] The predictor 163 based on the internal model is used to predict the optical / visual feedback changes that the disturbance should cause. Specifically, the predictor 163 based on the internal model implements the prediction function through a parallel processing mechanism: while sending the exploratory disturbance signal ΔI_k to the hardware, it is simultaneously input into the internal dynamic state space model; the model predicts the corresponding optical output change based on the current junction temperature, aging degree and other internal states; then, through a preset simplified "light-vision" conversion function (such as the luminous flux-image grayscale value mapping relationship), the optical prediction value is converted into the corresponding visual feedback change, forming a complete prediction closed loop.

[0095] Model optimizer 164 is used to calculate the error between predicted changes and actual feedback changes, and to update the parameters of the light source characteristic model unit using gradient descent. Specifically, model optimizer 164 achieves model self-calibration through an "evaluation-optimization" closed loop: First, within the synchronization time window of the injected perturbation signal ΔI_k, the actual visual feedback ΔF_actual_k is acquired, and the residual Error_k = ΔF_pred_k - ΔF_actual_k is calculated with the predicted value ΔF_pred_k; then, a loss function Loss = Σ(Error_k)^2 based on the least squares method is constructed to quantify the total model deviation; finally, through the gradient descent optimizer, the gradient ∇Loss is obtained by differentiating the loss function with respect to the key model parameters (state space matrix / nonlinear mapping table), and the parameters are iteratively updated according to θ_new = θ_old - learning_rate ∇Loss (learning_rate is the learning rate) until the loss function converges, completing the online calibration of the model parameters.

[0096] Example of implementation process: A light source controller deployed on a production line experienced a 10% light decay in its No. 3 LED channel after a year of intensive use. This process utilizes a closed-loop control mechanism of "measurement-comparison-correction" to achieve self-updating of the controller model. 1. Start Calibration: During production line maintenance breaks, engineers send the “START_CALIBRATION(channel=3)” command to the controller via the host computer.

[0097] 2. Triggering and initialization: After receiving the calibration command, the system stabilizes the channel to be calibrated at the reference working point (e.g., 500mA) and starts visual feedback synchronous acquisition.

[0098] 3. Active Exploration and Data Acquisition: The exploratory signal generator superimposes a series of tiny perturbations (e.g., ±5mA) onto the reference current. The system simultaneously performs two operations: Apply a disturbance: drive the LED with the disturbed current.

[0099] Parallel prediction: Input the same perturbation into the internal model and predict the resulting optical output (and the corresponding image grayscale change).

[0100] 4. Deviation Quantization: The system compares the predicted grayscale change under each perturbation with the grayscale change measured by the actual vision system, calculates the systematic prediction error, and constructs a loss function accordingly.

[0101] 5. Model parameter optimization: Optimize the algorithm to analyze the loss function, determine the correction direction and magnitude of key model parameters (such as photoelectric conversion efficiency), and perform a small iterative update.

[0102] 6. Convergence and Completion: Repeat steps 2 to 4, using multiple "explore-measure-update" cycles to continuously reduce model parameters (such as photoelectric conversion efficiency) by approximately 10%. When the model predictions and actual feedback re-align, the loss function converges to its minimum value, and the calibration process automatically terminates.

[0103] Therefore, through the collaborative work of the above-mentioned units and modules, this application transforms a multi-channel light source controller from a passive current output device into an intelligent optical generative system that can directly control light, autonomously coordinate, and self-evolve with the environment.

[0104] On the other hand, this application also provides a control method for the multi-channel light source controller.

[0105] See Figure 10 , Figure 10 This is a flowchart illustrating an embodiment of the control method for the light source controller provided in this application.

[0106] In some embodiments, the multi-channel light source control method provided in this application includes: S1, Instruction reception: If a non-real-time or soft real-time upper-layer instruction is received through the configuration instruction interface, proceed to steps S2 to S7. If an external synchronization signal requiring nanosecond or microsecond time determinism is received through the hard real-time trigger interface, proceed to steps S6 to S7. S2 executes multi-channel collaborative control and plans dynamic collaborative trajectories between channels; S3, based on the dynamic photoelectric and thermal characteristic model of the light source, solves the optical target in reverse to obtain the driving command; S4, perform feedforward compensation on the drive command based on the inter-channel coupling interference model; S5 performs safety verification and minimization correction on the compensated driving instructions based on multi-dimensional safety boundary functions, and generates safe final driving instructions. S6, based on the advanced timing control unit, generates high-precision, low-jitter timing drive signals according to the final drive command or directly received external synchronization signals; S7, using the generated timing drive signal to control multiple independent constant current drive channels.

[0107] In another embodiment, the main microcontroller 10 introduces a complex task execution engine 15, and based on this, provides another control method for strategically allocating tasks according to their complexity and real-time requirements. This execution strategy includes: for tasks requiring multi-channel dynamic coordination, the multi-channel coordination control unit is invoked according to the generated coordination task instructions, which generates the coordination trajectory and instantaneous target, and then outputs the final drive instruction to the advanced timing control unit through the joint collaboration of the light source characteristic model unit, the channel coupling compensation unit, and the intelligent safety protection unit; for basic, independent operations, the advanced timing control unit is directly scheduled according to its decomposed atomic operation sequence to efficiently and accurately execute the entire lighting sequence autonomously. For details, please refer to Figure 11 , Figure 11 This is a flowchart illustrating another embodiment of the control method for the light source controller provided in this application.

[0108] The method includes: S11 receives non-real-time or soft real-time upper-layer commands through the configuration command interface; S12, the complex task execution engine converts the received macro instructions into two types of lower-level instructions, for tasks requiring multiple passes. For dynamic collaborative task instructions, continue executing steps S13 to S18; for basic, independent operation instructions, continue executing steps S17 to S18. S13, executes multi-channel collaborative control, and plans dynamic collaborative trajectories between channels; S14, based on the dynamic photoelectric and thermal characteristic model of the light source, solves the optical target in reverse into driving commands; S15, perform feedforward compensation on the drive command based on the inter-channel coupling interference model; S16, Based on the multi-dimensional safety boundary function, the safety verification and minimization correction of the compensated driving instructions are performed to generate safe final driving instructions; S17, the advanced timing control unit generates high-precision, low-jitter timing drive signals according to instructions; S18, using the generated timing drive signal to control multiple independent constant current drive channels.

[0109] In summary, the above-mentioned technologies have brought unexpected technical effects to this application: By constructing a heterogeneous architecture of "software and hardware separation" and an intelligent algorithm with an embedded physical model, this invention elevates the multi-channel light source controller from a passive current execution device to an intelligent optical innovation system that can directly control light, actively predict interference, achieve precise coordination of dynamic processes, and autonomously learn and optimize throughout its entire life cycle. Ultimately, it simultaneously achieves complex functional flexibility, nanosecond-level synchronization, high security under access control performance, and long-term precision and stability on a single device, which are impossible to achieve with traditional technologies.

[0110] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A multi-channel light source controller, comprising a main microcontroller, multiple independent constant current drive channels, and a lead timing control unit physically independent of the main microcontroller, characterized in that: The input interface of the light source controller is configured into two types at the hardware level: A configuration instruction interface is connected to the main microcontroller to receive non-real-time or soft real-time upper-layer instructions. The hard real-time trigger interface is directly connected to the advanced timing control unit and is used to receive external synchronization signals that require nanosecond or microsecond time determinism. The main microcontroller integrates: The multi-channel collaborative control unit is used to uniformly plan an optimal path that strictly coordinates the entire dynamic trajectory from the starting point to the end point when multiple channels need to change synchronously based on upper-level instructions, and to generate instantaneous target values ​​that are coordinated in time. The light source characteristic model unit is used to establish a dynamic model describing the photoelectric and thermal characteristics of the light source load for each channel, and to solve the instantaneous target value into a driving command based on the model. The channel coupling compensation unit is used to predict the driving command based on a pre-established multi-input multi-output transfer function model that describes the physical coupling relationship between channels, and generate a feedforward compensation signal to suppress mutual interference between channels. The intelligent safety protection unit is used to predict the execution consequences of the compensated driving instructions based on a multi-dimensional safety boundary function, and to minimize the correction of the instructions when they may exceed the safe area, so as to generate safe final driving instructions. The lead timing control unit is configured to: The system receives the final drive command from the main microcontroller and generates high-precision, low-jitter timing drive signals. The system receives external synchronization signals from the hard real-time trigger interface and converts the external synchronization signals into high-precision, low-jitter timing drive signals independently of the main microcontroller. The multi-channel collaborative control unit includes: A synchronization task instruction parser, used to parse instructions containing target values, durations, and synchronization modes; A joint state space and trajectory planner is used to plan a trajectory within a joint state space consisting of channels participating in synchronization. A smooth trajectory from the current state to the target state; A real-time interpolator is used to sample on the trajectory to generate a sequence of instantaneous target values ​​for each control cycle; The instruction distribution module is used to distribute the instantaneous target value to the underlying controller of each channel.

2. The multi-channel light source controller according to claim 1, characterized in that, The light source characteristic model unit is configured to: iteratively solve a dynamic system model and inversely calculate the user-defined instantaneous target value output as the corresponding drive current command. This unit includes: A model parameter repository is used to pre-store a set of model parameters obtained through offline calibration and knowledge distillation for each supported LED model; The real-time state observer is used to maintain a current state vector for each channel in real time. It iteratively updates the internal state by acquiring the actual output current of the driving circuit and the temperature of the LED substrate in real time, combined with the internal thermal model and aging model. A forward predictor is used to receive a hypothetical drive current value and calculate the predicted optical parameters that the LED will produce under that hypothetical current. The reverse control command generator iteratively calls the forward predictor to solve for the drive current value that enables the predicted optical output to match the user-defined target.

3. The multi-channel light source controller according to claim 1, characterized in that, The number of the multiple independent constant current drive channels is eight.

4. The multi-channel light source controller according to claim 1, characterized in that, The channel coupling compensation unit includes: A multi-input multi-output coupled disturbance transfer function model, which is obtained through system identification during the controller design or production calibration phase, is used to describe the voltage disturbance caused by the current change of any channel to the other channels; The drive instruction preprocessor is used to calculate the expected current change rate of each channel drive instruction; A real-time disturbance predictor is used to predict the equivalent voltage disturbance that all channels will experience based on the transfer function model and the expected rate of change of current. The compensation signal injection module is used to calculate the feedforward compensation value to be applied to each channel control loop based on the predicted voltage disturbance and the characteristics of the constant current drive circuit, and synthesize the compensation value with the original drive command to generate the final drive command with anti-interference.

5. The multi-channel light source controller according to claim 1, characterized in that, The intelligent security protection unit includes: The multidimensional safety boundary function definition module is used to define multiple state variables, including current, voltage, temperature, and power change rate. A safe zone has been formed; The system state prediction module is used to predict the future system state after executing the compensated drive command based on the light source characteristic model unit. The hazard assessment module, when it predicts that the future system state may exceed the safe zone, initiates an optimization process to calculate a correction amount that minimizes changes to the original instructions; The online optimizer, when risks exist, solves a problem that aims to minimize instruction corrections while satisfying safety boundaries. The optimization of the bundle is addressed, and the corrected safety instructions are output.

6. The multi-channel light source controller according to claim 1, characterized in that, The advanced timing control unit is a field-programmable gate array (FPGA) or a complex programmable logic device (CPLD), which internally includes: The input signal conditioning and arbitration module is used to electrically isolate external trigger signals through optocouplers, and to filter out high-frequency noise and mechanical jitter of the input signal through hardware-implemented digital low-pass filters or debouncing logic, and to capture the high-precision timestamp of each valid trigger signal edge using a high-frequency internal clock. The timing instruction decoding and parameter register group is used to receive and store timing task configurations from the main microcontroller; The atomic timing operation library solidifies the basic, indivisible atomic timing operation logic; The high-precision timing sequence generator, based on the atomic timing operation library and configuration parameters, autonomously and accurately generates the final underlying drive control waveform using a high-frequency internal clock and multiple high-precision counters.

7. The multi-channel light source controller according to any one of claims 1 to 6, characterized in that, The main microcontroller also includes a complex task execution engine, which receives and parses non-real-time or soft real-time upper-layer macro instructions and autonomously decomposes them into execution strategies based on an internally stored task knowledge base; the execution strategies include: For tasks requiring multi-channel dynamic coordination, the multi-channel coordination control unit is invoked according to the generated coordination task instructions. The control unit generates the coordination trajectory and instantaneous target, and then outputs the final drive instruction to the advanced timing control unit through the joint cooperation of the light source characteristic model unit, channel coupling compensation unit and intelligent safety protection unit. For basic, independent operations, the advance timing control unit is directly scheduled based on the sequence of atomic operations it decomposes, so as to execute the entire lighting sequence autonomously with high efficiency and precision. The complex task execution engine includes: The task knowledge base stores the definitions of multiple complex lighting tasks and their scripts decomposed into atomic operations; An instruction parser and decomposer is used to receive macro instructions, retrieve corresponding tasks from the knowledge base, and generate an atomic operation queue. The atomic lighting operation library contains optimized low-level control functions for real-time scheduling executors to call; A real-time scheduler is used to retrieve and execute instructions in the atomic operation queue in sequence.

8. The multi-channel light source controller according to any one of claims 1 to 6, characterized in that, It also includes an online self-calibration module, which combines feedback signals from the external vision system with predicted values ​​from the internal light source characteristic model unit, and automatically updates model parameters through optimization algorithms to compensate for performance drift caused by aging or environmental changes in the light source load. The online self-calibration module includes: An external feedback interface is used to receive image quality feedback signals from the vision system; An exploratory signal generator is used to apply small driving perturbations near the stable operating point; An internal model-based predictor is used to predict the optical / visual feedback changes that the disturbance should cause; The model optimizer calculates the error between the predicted change and the actual feedback change, and updates the model using gradient descent. Parameters of the light source characteristic model unit.

9. A multi-channel light source control method, applied to a multi-channel light source controller as described in any one of claims 1 to 6, characterized in that, The method includes: S1, Instruction reception: If a non-real-time or soft real-time upper-layer instruction is received through the configuration instruction interface, proceed to steps S2 to S7. If an external synchronization signal requiring nanosecond or microsecond time determinism is received through the hard real-time trigger interface, proceed to steps S6 to S7. S2 executes multi-channel collaborative control and plans dynamic collaborative trajectories between channels; S3, based on the dynamic photoelectric and thermal characteristic model of the light source, solves the optical target in reverse to obtain the driving command; S4, perform feedforward compensation on the drive command based on the inter-channel coupling interference model; S5 performs safety verification and minimization correction on the compensated driving instructions based on multi-dimensional safety boundary functions, and generates safe final driving instructions. S6, based on the advanced timing control unit, generates high-precision, low-jitter timing drive signals according to the final drive command or directly received external synchronization signals; S7, using the generated timing drive signal to control multiple independent constant current drive channels.

10. A multi-channel light source control method, applied to the multi-channel light source controller as described in claim 7, characterized in that, The method includes: S11 receives non-real-time or soft real-time upper-layer commands through the configuration command interface; S12, the complex task execution engine converts the received macro instructions into two types of lower-level instructions. For task instructions that require multi-channel dynamic coordination, steps S13 to S18 are executed; for basic, independent operation instructions, steps S17 to S18 are executed. S13, executes multi-channel collaborative control, and plans dynamic collaborative trajectories between channels; S14, based on the dynamic photoelectric and thermal characteristic model of the light source, solves the optical target in reverse into driving commands; S15, perform feedforward compensation on the drive command based on the inter-channel coupling interference model; S16, Based on the multi-dimensional safety boundary function, the safety verification and minimization correction of the compensated driving instructions are performed to generate safe final driving instructions; S17, the advanced timing control unit generates high-precision, low-jitter timing drive signals according to instructions; S18, using the generated timing drive signal to control multiple independent constant current drive channels.

Images