Remote parameter configuration method for LED driving power supply

Abstract

The invention relates to the technical field of digital power supply control, in particular to a remote parameter configuration method for an LED driving power supply, which is applied to a series lighting cluster and comprises the following steps: a head node analyzes a configuration packet to generate a dynamic token embedded with dual variables and a power accumulator and issues the dynamic token along a link; the target node captures the dynamic token, collects the local environment temperature and the AC voltage, deduces the thermal power and the power grid critical safety power respectively, and takes an intersection to determine a local feasible region; the nominal expected power is combined with dual variables to project to a local feasible region to obtain local distribution power, and dynamic tokens are accumulated and updated and transmitted to a next node; the tail node calculates the power profit and loss of the whole network, updates dual variables and judges a convergence state; if convergent, each node converts the local distribution power into a voltage value to execute physical driving, and if not convergent, the dynamic token is returned to deduce again. According to the invention, adaptive compensation of cluster power is realized, and the problem of thermal runaway of a single machine or unbalance of global luminous flux is avoided.

Description

Technical Field

[0001] This invention relates to the field of digital power control technology, and in particular to a method for remote parameter configuration of an LED driver power supply. Background Technology

[0002] In large-scale, high-density plant factories with multi-unit cascaded supplemental lighting, the physical space suffers from severe heterogeneity in heat dissipation and long-distance power supply line losses. When the host computer issues a full-load configuration command to the lighting cluster, LED drivers located in ventilation dead zones are prone to exceeding physical heat dissipation limits and thermal breakdown of power devices if forced to execute the full-load output command. At the same time, the simultaneous surge in output current from a large number of high-power devices can cause a transient and steep drop in local AC grid voltage.

[0003] Traditional parameter configuration architectures mostly employ fixed parameters distributed uniformly, coupled with independent overload protection for each individual machine. Under these architectures, relying solely on independent over-temperature or under-voltage protection mechanisms at the underlying hardware level to enforce derating, while preventing hardware damage, can lead to severe localized light deficits within the planting area. Furthermore, the lack of cross-physical node computing power collaboration and power compensation mechanisms in traditional parameter configuration architectures creates an irreconcilable technical contradiction between individual hardware self-protection and a constant global quantum flux density distribution. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a remote parameter configuration method for LED driver power supplies, aiming to improve the problem that traditional parameter configuration mostly uses fixed parameters to be uniformly issued, which ignores the physical heat dissipation of nodes and the differences in the power grid, thus causing individual units to be prone to thermal runaway or global luminous flux imbalance.

[0005] This invention provides the following technical solution: a remote parameter configuration method for an LED driver power supply, applied to a series lighting cluster, wherein each node is configured with cascaded analysis, boundary deduction, and drive execution modules, including the following steps: S1. The cascaded parsing module of the first node parses the configuration package to obtain the nominal expected power, generates a dynamic token with embedded dual variables and power accumulator, and distributes it along the link; S2. When the boundary deduction module of the target node captures the issued dynamic token, it collects the local ambient temperature and AC voltage, and respectively deduces the thermal and grid critical safety power and takes the intersection to establish the local feasible region. S3. The target node's drive execution module extracts the dual variable from the dynamic token, projects the parsed nominal expected power into the established local feasible region in combination with the dual variable to obtain the local allocated power, and accumulates the local allocated power into the power accumulator to update the dynamic token and pass it to the next node. S4. When the updated dynamic token is transferred to the tail node, the tail node calculates the power surplus and deficit of the entire network based on the power accumulator that has been accumulated, updates the dual variable and determines whether convergence has occurred. S5. If convergence is determined, each drive execution module will convert the projected local allocated power into a voltage value and write it into a register to execute the physical drive; if convergence is determined not to be achieved, the dynamic token containing the new dual variable will be sent back to the first node to re-trigger the deduction process.

[0006] By adopting the above technical solution, projection allocation is performed by combining dynamic tokens with local feasible domains, thereby realizing adaptive compensation of cluster power across nodes. This improves the problem that traditional parameter configuration mostly uses fixed parameters to be uniformly distributed, which ignores the physical heat dissipation of nodes and differences in the power grid, resulting in individual machines being prone to thermal runaway or global optical flux imbalance.

[0007] Optionally, in S1, the cascaded parsing module of the first node parses the configuration packet to obtain the nominal expected power, including: The cascading parsing module of the first node receives communication data streams through a network interface; The communication data stream is unpacked and subjected to cyclic redundancy check to extract the global expected total power; Send topology probe frames to obtain the total number of online LED driver power supplies; The nominal expected power is obtained by dividing the global expected total power by the total number of LED driver power supplies.

[0008] Optionally, in S1, generating a dynamic token with an embedded dual variable and a power accumulator and distributing it along the link includes: Allocate contiguous blocks of data in the microcontroller's memory; The address spaces corresponding to the dual variables and the power accumulator in the data block are cleared and initialized; Add a frame header identifier and the target routing address to the data block and encapsulate it into the dynamic token; The dynamic token is pushed into the send buffer of the serial communication interface for distribution.

[0009] Optionally, in S2, the separate deduction of thermal and grid critical safety power includes: Read the analog-to-digital conversion sample value of the external thermistor; The maximum driving current allowed to pass under the current heat dissipation boundary is calculated by calling the preset steady-state thermal resistance matrix, and the maximum driving current is mapped to the thermal critical safe power. The transient sag slope is calculated by acquiring the zero-crossing time and peak value data of the AC voltage. By combining the impedance factor of the power supply phase and the transient drop slope, the critical safe power of the power grid to prevent AC bus undervoltage can be deduced.

[0010] Optionally, in S2, the step of taking the intersection to establish the local feasible region includes: A numerical comparison calculation is performed between the derived thermal critical safety power and the power grid critical safety power; Choose the smaller value as the upper limit of the absolute safe power; The minimum dimming power required to maintain continuous oscillation of the internal switching power supply chip is taken as the absolute safe power lower bound. The lower bound of the absolute safe power and the upper bound of the absolute safe power are combined to construct a continuous mathematical closed interval, forming the local feasible region.

[0011] Optionally, in S3, the step of projecting the analytically obtained nominal expected power into the established local feasible region in combination with the dual variable to obtain the local allocated power includes: The theoretical compensatory power is obtained by calculating the difference between the nominal expected power and the dual variable; Perform a one-dimensional Euclidean projection operation to compare the theoretical compensation power with the upper bound and the lower bound of the absolute safe power of the local feasible region. The theoretical compensatory power is truncated under the constraint of not exceeding the upper and lower bounds of the absolute safe power, and the local allocated power is output.

[0012] Optionally, in S3, accumulating the locally allocated power to the power accumulator to update the dynamic token and passing it to the next node includes: Extract the old power accumulation value written by the previous node from the payload segment of the captured dynamic token; The calculated local power allocation is added to the old power accumulation value to obtain a new power accumulation value; The new power accumulation value is overwritten into the payload segment of the dynamic token; The updated dynamic token is sent to the next node in the physical topology via the transmit pin of the serial communication interface.

[0013] Optionally, in S4, the tail node calculates the power surplus / deficit of the entire network based on the power accumulator after accumulation, and updates the dual variable including: Extract the preset subgradient iteration step size parameter; The final new power accumulation value in the dynamic token flowing to the tail node is subtracted from the global expected total power to obtain the overall network power surplus or deficit. The step bias is obtained by multiplying the total network power surplus / deficit with the subgradient iteration step size parameter. The update operation is completed by adding the step bias to the current dual variable.

[0014] Optionally, in S4, determining whether convergence has occurred includes: Calculate the absolute difference between the updated dual variable and the original dual variable; Extract the convergence threshold parameters that are pre-stored in non-volatile memory; Compare the absolute difference with the convergence threshold parameter. If the absolute difference is less than or equal to the convergence threshold parameter, output a Boolean truth value indicating convergence; If the absolute difference is greater than the convergence threshold parameter, output a Boolean false value indicating that convergence has not been achieved.

[0015] Optionally, in S5, the process of each drive execution module converting the projected local allocated power into a voltage value and writing it into a register to execute the physical drive includes: Receive the convergence lock command broadcast by the tail node; The microcontroller's internal digital-to-analog conversion mapping table is invoked to convert the locally allocated power retained in the last iteration into the corresponding target reference voltage digital value; Write the target reference voltage digital value into the control register of the digital-to-analog converter; The analog voltage is output to the feedback pin of the pulse width modulation controller to adjust the duty cycle of the power switch.

[0016] The present invention has the following beneficial effects: 1. In this invention, dynamic tokens are combined with local feasible domains for projection allocation, thereby realizing cross-node adaptive compensation of cluster power. This improves the problem that traditional parameter configuration mostly uses fixed parameters for unified distribution, which ignores the physical heat dissipation of nodes and differences in the power grid, resulting in a single machine being prone to thermal runaway or global optical flux imbalance.

[0017] 2. In this invention, the local feasible region is established by simultaneously deriving the thermal and grid critical safe power and taking the intersection, thereby accurately locking the absolute safe output range of a single unit. This improves the problem that traditional overload protection mostly adopts a single passive over-temperature voltage reduction, which lacks multi-dimensional physical limit prediction, thus causing the equipment to be easily damaged by thermal shock.

[0018] 3. In this invention, the theoretical compensation power is projected and truncated onto the local feasible region, thereby strictly controlling the upper limit of node power allocation. This improves the problem that traditional cluster control mostly adopts host-forced computing power allocation, which lacks hard adjudication of the underlying physical boundary, resulting in over-limit output under harsh working conditions.

[0019] 4. In this invention, the tail node performs subgradient descent updates of the dual variables based on the power surplus and deficit of the entire network, thereby guiding the computing power of the entire network toward the global optimal allocation. This improves the problem that traditional multi-machine collaboration mostly uses external central servers for computing, which is heavily dependent on high-bandwidth communication networks, resulting in high system deployment costs. Attached Figure Description

[0020] Figure 1 This is a flowchart of a remote parameter configuration method for an LED driver power supply proposed in this invention; Figure 2 This is a flowchart illustrating the configuration distribution and local feasible domain establishment of a remote parameter configuration method for an LED driver power supply proposed in this invention. Figure 3 This is a flowchart of the local optimization projection and dynamic token update process for a remote parameter configuration method for an LED driver power supply proposed in this invention. Figure 4 This is a flowchart illustrating the global profit and loss determination and physical drive execution of a remote parameter configuration method for an LED driver power supply proposed in this invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0022] In a first embodiment of the present invention, a remote parameter configuration method for an LED driver power supply is provided, applied to a series lighting cluster, wherein each node is configured with cascaded analysis, boundary deduction, and drive execution modules, such as... Figures 1-4 As shown, it includes the following steps: S1. The cascaded parsing module of the first node parses the configuration package to obtain the nominal expected power, generates a dynamic token with embedded dual variables and power accumulator, and distributes it along the link; Furthermore, in S1, the cascaded parsing module of the first node parses the configuration package to obtain the nominal expected power, which includes: The cascading parsing module of the first node receives communication data streams through the network interface; The communication data stream is unpacked and cyclic redundancy check is performed to extract the global expected total power; Send topology probe frames to obtain the total number of online LED driver power supplies; The nominal expected power is obtained by dividing the total expected global power by the total number of LED driver power supplies.

[0023] In S1, generating a dynamic token with an embedded dual variable and a power accumulator and distributing it along the link includes: Allocate contiguous blocks of data in the microcontroller's memory; The address space corresponding to the even variables and power accumulator in the data block is cleared and initialized; Add a frame header identifier and the target route address to the data block and encapsulate it into a dynamic token; The dynamic token is pushed into the send buffer of the serial communication interface for distribution.

[0024] Specifically, step S1, as the initialization starting point for the remote parameter configuration of the entire serial lighting cluster, plays a core role in instruction conversion and the construction of the underlying communication link. The cascading parsing module of the first node acts as a gateway between the host computer and the local cluster, completing the mapping from macroscopic system instructions to microscopic node operation variables.

[0025] The data input is the raw communication data stream sent by the host computer over the network. After receiving this data stream, the first node performs unpacking and cyclic redundancy check (CRC) operations, stripping away the additional envelope of the communication protocol and filtering out bit errors generated in the transmission link. After the check passes, the microcontroller extracts the global expected total power within the command payload. To determine the actual carrying capacity of the current physical network, the first node actively sends topology probe frames to the subsequent links, accurately counting the total number of normally connected and online LED driver power supplies using the response mechanism of each node.

[0026] Based on the extracted total power and the detected number of online nodes, the first node's arithmetic logic unit performs a division operation to establish a network-wide baseline reference value. The calculation formula is as follows: ; in This represents the nominal expected power of each online LED driver. This represents the global expected total power as specified by the host computer instructions. This represents the total number of LED drivers actually detected as being online. This calculation logic accurately distributes the massive system-level illuminance requirements to currently healthy and available physical nodes, preventing technical issues caused by blindly issuing fixed parameters due to power outages or disconnections of individual slave devices, which could lead to insufficient total quantum flux density across the entire network.

[0027] In the process of generating the messenger carrier required for the optimization algorithm, the microcontroller allocates a contiguous block of data using a dynamic memory allocation mechanism. The cascaded parsing module performs a zeroing initialization operation on the dedicated address space within this data block to establish the initial values ​​for the algorithm's iterations. The mathematical expression for variable initialization is as follows: ; in The dual variable represents the initial state. Setting it to zero means that there is no optical flux compensation bias in the system before the iterative optimization begins. This represents the initial value of the power accumulator parameters for the entire network, used to reserve blank accumulation space for power reporting by subsequent nodes at all levels.

[0028] After variable initialization, the microcontroller adds the frame header identifier and target routing address required for underlying communication to the above data block, packaging it into a dynamic token with network transmission capability. Finally, it triggers a low-level hardware interrupt, pushing the assembled dynamic token into the transmit buffer of the serial communication interface, and outputting it to the next cascaded device via the transmit pin.

[0029] From the perspective of data output flow, this step ultimately produces two core results. The first is the nominal expected power, which is stored in the local memory of the first node and serves as a benchmark for subsequent optimization by the first node. The second is a dynamic token that flows outward via the serial bus. This token carries a clean dual variable and an empty power accumulator, officially initiating the data transmission process for the distributed computing across the entire network.

[0030] S2. When the boundary deduction module of the target node captures the issued dynamic token, it collects the local ambient temperature and AC voltage, and respectively deduces the thermal and grid critical safety power and takes the intersection to establish the local feasible region. Furthermore, in S2, the thermal and grid critical safety power are respectively derived, including: Read the analog-to-digital conversion sample value of the external thermistor; The maximum driving current allowed under the current heat dissipation boundary is calculated by calling the preset steady-state thermal resistance matrix, and the maximum driving current is mapped to the thermal critical safe power. Calculate the transient sag slope by obtaining the zero-crossing time and peak value data of the AC voltage; By combining the impedance factor and transient drop slope of the power supply phase, the critical safe power of the power grid to prevent AC bus undervoltage is deduced.

[0031] In S2, determining the local feasible region by taking the intersection includes: A numerical comparison calculation is performed between the derived thermal critical safe power and the grid critical safe power. Choose the smaller value as the upper limit of the absolute safe power; The minimum dimming power required to maintain continuous oscillation of the internal switching power supply chip is taken as the absolute safe power lower bound. The lower bound and upper bound of the absolute safe power are combined to construct a continuous mathematical closed interval, forming a local feasible region.

[0032] Specifically, step S2, as the local constraint generation step of the distributed cooperative control algorithm, plays a core role in transforming the complex physical environment of the LED driver power supply into mathematically rigid boundary conditions, thereby preventing a single device from blindly executing high-power commands under extreme conditions, which could lead to hardware damage or power grid collapse.

[0033] The boundary deduction module first performs a safety assessment from a thermodynamic perspective. The data input terminal acquires the analog-to-digital conversion sample value of the external thermistor, which the microcontroller interprets as the local ambient temperature. The system then calls a preset steady-state thermal resistance matrix, combines it with the current heat dissipation boundary to calculate the maximum permissible drive current, and maps it to the thermally critical safe power. This deduction process follows the steady-state heat conduction calculation formula: ; in Represents the thermally critical safe power. This represents the preset limit junction temperature allowed by the power device. This represents the local ambient temperature calculated from the sampled value of an external thermistor. This represents the combined thermal resistance from the device to the environment, as defined in the steady-state thermal resistance matrix. This calculation logic establishes the maximum power output of the device without thermal penetration.

[0034] Regarding safety constraints in the electrical dimension of the power grid, the microcontroller's sampling port continuously acquires the zero-crossing time and peak value data of the AC voltage, and performs differential calculations on the time axis to obtain the transient sag slope. The system combines the inherent impedance factor of the power supply phase where the equipment is located with this transient sag slope to deduce the critical safe power of the power grid to prevent AC bus undervoltage. The derivation model expression is as follows: ; in Represents the critical safe power of the power grid. This represents the maximum allowable AC voltage sag slope threshold set by the system. This represents the actual transient drop slope calculated based on the zero-crossing time and peak data. This represents the impedance factor of the AC power supply phase where the target node is located. This calculation blocks the path for the device to continue drawing large currents when grid fluctuations intensify.

[0035] After the calculations in the two orthogonal dimensions described above, the data flow enters the intersection decision stage. The arithmetic logic unit executes a numerical comparison instruction on the derived thermal critical safety power and the grid critical safety power, strictly selecting the smaller of the two values ​​as the upper bound of the absolute safety power. The system also reads the minimum dimming power required to maintain the continuous oscillation of the internal switching power supply chip, and uses this as the lower bound of the absolute safety power.

[0036] The microcontroller ultimately combines the lower and upper bounds of the absolute safe power to construct a continuous mathematically closed interval, which is then output as the local feasible region. The expression for this mathematically closed interval is as follows: ; in This represents the finally established local feasible region. This represents the minimum dimming power required to maintain normal hardware operation, i.e., the lower limit of the absolute safe power. This represents the function that takes the minimum value. Thus, step S2 completely transforms the externally acquired discrete temperature and voltage data into the numerical constraint space necessary for the next optimization algorithm.

[0037] S3. The target node's drive execution module extracts the dual variable from the dynamic token, combines the parsed nominal expected power with the dual variable, projects it onto the established local feasible region to obtain the local allocated power, accumulates the local allocated power to the power accumulator to update the dynamic token and passes it to the next node. Furthermore, in S3, the analytically obtained nominal expected power is projected onto the established local feasible region in conjunction with the dual variables to obtain the locally allocated power, which includes: The theoretical compensatory power is obtained by calculating the difference between the nominal expected power and the dual variable; Perform a one-dimensional Euclidean projection operation to compare the theoretical compensation power with the upper and lower bounds of the absolute safe power in the local feasible region. The theoretical compensatory power is truncated under the constraint of not exceeding the upper and lower limits of the absolute safe power, and the local distributed power is output.

[0038] In S3, accumulating locally allocated power into the power accumulator to update the dynamic token and passing it to the next node includes: Extract the old power accumulation value written by the previous node from the payload segment of the captured dynamic token; The calculated local power allocation value is added to the old power accumulation value to obtain the new power accumulation value; Write the new power accumulation value into the payload segment of the dynamic token; The updated dynamic token is sent to the next node in the physical topology via the transmit pin of the serial communication interface.

[0039] Specifically, step S3, which receives macroscopic instructions from the system and microscopic environmental parameters from the nodes, is the core computational step for the cluster system to complete distributed computing power allocation. The target node's driver execution module captures the dynamic token from the network data stream and extracts the dual variables and accumulated old power values ​​contained within it. The microcontroller uses the nominal expected power pre-stored in memory, along with the upper and lower bounds of the absolute safe power established in the preceding steps, as the core computational inputs.

[0040] After entering the computational solution phase, the arithmetic logic unit executes a subtraction instruction to calculate the difference between the nominal expected power and the dual variable, thus obtaining the theoretical compensating power. The specific formula expression is as follows: ; in Represents the theoretical compensation power. The nominal expected power obtained from the analysis, This represents the dual variable extracted from the dynamic token. The system uses this formula to add the power share missing globally due to limitations in some nodes to the current node's baseline calculation.

[0041] The microcontroller then performs a one-dimensional Euclidean projection calculation. The system uses the absolute safety power upper and lower bounds as absolute constraints to perform a hard numerical truncation of the theoretical compensation power. The logic of this projection calculation satisfies the following formula: ; in This represents the locally distributed power output after projection. Represents the lower bound of absolute safe power. Represents the upper limit of absolute safe power. and These refer to the operations of finding the maximum and minimum values, respectively. This processing mechanism directly eliminates the possibility of operations exceeding the limits, forcing the compensatory instructions, which theoretically require high output, to be reduced to a safe range that the local physical devices can withstand.

[0042] After the local power allocation calculation is completed, the driver execution module initiates the token payload update process. The microcontroller reads the residual old power accumulation value within the payload segment and adds it to the calculated local power. The addition formula is defined as follows: ; in This represents the newly accumulated power value generated by the merger. The previous node writes the old power accumulation value. The arithmetic logic unit overwrites the new power accumulation value into the designated memory area of ​​the dynamic token. The underlying hardware interface chip then extracts the reconstructed payload data and forwards it to the next connected node via the transmit pin of the serial communication interface, following the physical topology harness.

[0043] Based on the data output flow, this step generates two independent branches. The first branch is the locally allocated power temporarily stored in the microcontroller's local memory, serving as a reference for subsequent constant current output. The second branch is a dynamic token carrying the newly accumulated power value, which continues to flow along the link. This process, without relying on central server scheduling, independently completes the best-effort power distribution within the safety threshold using only the computing power of a single edge machine, providing a reliable data source for subsequent tail nodes to perform global profit and loss determination.

[0044] S4. When the updated dynamic token is transferred to the tail node, the tail node calculates the power surplus and deficit of the entire network based on the power accumulator that has been accumulated, updates the dual variables and determines whether convergence has occurred. Furthermore, in S4, the tail node calculates the power surplus / deficit of the entire network based on the completed power accumulator, and updates the dual variables including: Extract the preset subgradient iteration step size parameter; The final new power accumulation value in the dynamic tokens flowing to the tail node is subtracted from the global expected total power to obtain the power surplus or deficit of the entire network. The step bias is obtained by multiplying the total network power surplus / deficit with the subgradient iteration step size parameter. The update operation is completed by adding the step bias to the current dual variable.

[0045] In S4, determining whether convergence has occurred includes: Calculate the absolute difference between the updated dual variable and the original dual variable; Extract the convergence threshold parameters that are pre-stored in non-volatile memory; Compare the absolute difference with the convergence threshold parameter. If the absolute difference is less than or equal to the convergence threshold parameter, output a Boolean truth value indicating convergence; If the absolute difference is greater than the convergence threshold parameter, output a Boolean false value indicating that the convergence has not been achieved.

[0046] Specifically, step S4 constitutes the feedback and decision-making stage of the entire distributed computing power allocation algorithm. The tail node at this stage undertakes the task of system-level error assessment, summarizing the global state based on the unidirectional flow of the physical topology, thereby determining whether the computing power allocation process should be terminated.

[0047] Regarding the data input and the underlying mathematical update mechanism, the dynamic token flows along the communication link to the tail node. Its internal payload carries the final new power accumulation value generated by the sequential accumulation of all nodes in the network. The microcontroller of the tail node takes over this data and retrieves the preset subgradient iteration step size parameter from the local storage area.

[0048] The arithmetic logic unit first executes the subtraction instruction, subtracting the final accumulated power value from the globally expected total power. The difference is the current power surplus or deficit for the entire network. The specific power surplus or deficit calculation formula is as follows: ; in Represents the overall power surplus or deficit of the network. This represents the final accumulated new power value in the dynamic tokens that flow to the tail node. This represents the total expected power globally. This step quantifies the gap between the total power that the entire network can actually allocate and the target total power set by the host computer.

[0049] Next, the microcontroller uses a multiplier to multiply the total network power surplus / deficit by the subgradient iteration step size parameter, generating a step bias. The system then directly adds this step bias to the current dual variable, thus completing the algebraic update of the dual variable. The mathematical expression for the overall update operation is as follows: ; in Represents the updated dual variable. Represents the current dual variable before the update. This represents the preset subgradient iteration step size parameter. This represents the calculated power surplus or deficit of the entire network. This operational logic establishes the system's automatic adjustment mechanism: when the power allocated to the entire network is insufficient, it guides the dual variable to shift in the direction of stimulating nodes to increase power; when the power allocated to the entire network is excessive, it suppresses the output tendency of the nodes in the opposite direction.

[0050] After the dual variable is updated, the data flow enters the convergence decision branch. The tail node uses the absolute value calculation rule to calculate the absolute difference between the updated and unupdated dual variables. The formula for calculating the difference is: ; in This represents the absolute difference between the old and new dual variables. Represents the updated dual variable. Represents the dual variable before the update. This represents the absolute value operation.

[0051] The microcontroller then addresses and accesses the non-volatile memory to retrieve the convergence threshold parameter, which was pre-written during device manufacturing or system initialization. The comparator circuit is activated, performing a hard comparison between the calculated absolute difference and the convergence threshold parameter.

[0052] Based on the comparison results, the tail node's general-purpose input / output ports or internal status registers will generate explicit data outputs. If the absolute difference is less than or equal to the convergence threshold parameter, the system determines that the power allocation scheme of all nodes in the network has reached the optimal solution under the safety boundary, and then outputs a Boolean true value indicating convergence to the algorithm's main control program. Conversely, if the absolute difference is greater than the convergence threshold parameter, it indicates that the network's computing power is still oscillating and searching for an equilibrium point, and the system outputs a Boolean false value indicating non-convergence. This Boolean state variable serves as the sole control beacon, directly instructing the subsequent system whether to execute physical emission actions or initiate the next round of token return iterations.

[0053] S5. If convergence is determined, each driver execution module will convert the projected local allocated power into a voltage value and write it into a register to execute the physical drive; if convergence is determined, the dynamic token containing the new dual variable will be sent back to the first node to re-trigger the deduction process. Furthermore, in S5, each driver execution module converts the projected locally allocated power into a voltage value and writes it to a register to execute the physical drive, including: Receive the convergence lock command broadcast by the tail node; The microcontroller's internal digital-to-analog conversion mapping table is invoked to convert the locally allocated power retained in the last iteration into the corresponding target reference voltage digital value; Write the target reference voltage digital value into the control register of the digital-to-analog converter; The analog voltage is output to the feedback pin of the pulse width modulation controller to adjust the duty cycle of the power switch.

[0054] Specifically, step S5 is the final step in the transformation of the entire distributed computing power allocation algorithm into the underlying hardware execution. This step converts the optimal power value at the mathematical level into control signals for the specific circuit, establishing a control closed loop from the software algorithm to physical light emission.

[0055] The data flow is triggered by the convergence determination of the difference sequence of the dual variables. If convergence is not determined, the dynamic token carries the new dual variable and is transmitted back to the first node via the serial bus, and the system restarts a new round of boundary inference and projection calculations. If convergence is determined, the tail node generates a convergence lock command and broadcasts it to the entire network. After receiving the command, the microcontroller communication ports of each target node in the cluster immediately freeze the locally allocated power reserved in the last iteration in their internal memory, using it as the absolute reference for the hardware constant current output.

[0056] Internally, the microcontroller retrieves a digital-to-analog conversion mapping table pre-stored in non-volatile memory. The arithmetic logic unit performs table lookup and linear interpolation operations to convert the previously frozen local power allocation into the corresponding target reference voltage digital value. The mathematical expression of this mapping conversion relationship is as follows: ; in The digital value representing the target reference voltage. Represents the scaling factor for digital-to-analog conversion. This represents the locally allocated power retained from the last iteration. This represents the nominal output voltage of the LED driver circuit. This represents the inherent zero-point offset constant of the hardware circuit. This formula scales the power dimension proportionally and fits it to a discrete digital quantity that the microcontroller can recognize.

[0057] Upon entering the physical output data stream stage, the microprocessor writes the target reference voltage digital value into the control register of the digital-to-analog converter (DAC) via the internal data bus. Based on the value stored in the register, the peripheral circuitry of the DAC generates a specific analog voltage at the physical pin and outputs this analog voltage directly to the feedback pin of the pulse width modulation (PWM) controller. The comparator circuit inside the PWM controller receives this analog voltage and continuously adjusts the duty cycle of the external power switch. The specific physical adjustment formula for the duty cycle satisfies: ; in This represents the duty cycle of the power switch. This represents the analog voltage output from the digital-to-analog converter to the feedback pin. This represents the peak voltage of the reference triangular wave generated by the internal oscillator of the pulse width modulation controller.

[0058] Based on the aforementioned underlying hardware action sequence, the power allocation command in the digital domain is directly translated into the conduction time of the power electronic devices. The entire physical execution process breaks down the electrical isolation between the network communication layer and the underlying driver circuit, ensuring that the actual luminous flux emitted by the LED light source strictly follows the safety configuration parameters deduced collaboratively by the preceding nodes, and implementing cluster heat penetration prevention and total luminous flux compensation actions at the hardware level. Example

[0059] In large-scale, high-density plant factories with multi-machine cascaded supplemental lighting, the three-dimensional planting architecture leads to severe heterogeneity in the physical space's heat dissipation environment and long-distance power supply line losses. When the host computer uniformly issues a full-load configuration command to the lighting cluster, LED driver power supplies located in ventilation dead zones are prone to exceeding physical heat dissipation limits and thermal breakdown of power devices if forced to execute the full-load output command. Simultaneously, the synchronous surge in output current of numerous high-power devices within a microsecond time window can cause transient voltage drops in the local AC grid, even tripping the distribution cabinet. While relying solely on independent over-temperature or under-voltage protection mechanisms at the underlying hardware level to enforce forced derating avoids hardware damage, it results in severe localized light deficiency within the planting area. Traditional remote configuration architectures, lacking cross-physical node computing power collaboration and power compensation mechanisms, cannot resolve the technical contradiction between single-machine hardware self-protection and a constant global quantum flux density distribution. To address these issues, this invention provides a remote parameter configuration method for LED driver power supplies, the structure of which is as follows... Figure 1 As shown. The specific implementation process of this method is as follows: The first node initiates communication by issuing a dynamic token with embedded global variables. The target node, by collecting real-time ambient temperature and AC voltage, deduces the local feasible region limited by hardware heat dissipation and power supply network limits, thus truncating the execution path that could lead to local overheating or grid undervoltage due to blindly responding to full-load commands. Nodes rely on extracted dual variables to project and truncate the local feasible region, deriving a safe local power allocation and accumulating their local computing power contribution into the token payload for transmission to subsequent stages. The tail node determines the overall profit and loss based on the converged global power data, driving the entire network to perform multiple rounds of deduction and trial calculations by transmitting updated dual variables until the convergence condition is met. Once this is achieved, each node synchronously maps digital commands directly to the underlying reference voltage signal to drive the power devices to emit light. The entire process relies solely on the physical state interaction of edge devices, achieving adaptive compensation allocation of the cluster's total power among nodes while ensuring absolute electrical and thermodynamic safety for individual machines. This eliminates the technical defect of unbalanced global luminous flux distribution caused by independent derating of devices in large-scale special lighting scenarios.

[0060] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A remote parameter configuration method for an LED driver power supply, applied to a series lighting cluster, wherein each node is configured with a cascaded analysis, boundary deduction, and drive execution module, characterized in that, Includes the following steps: S1. The cascaded parsing module of the first node parses the configuration package to obtain the nominal expected power, generates a dynamic token with embedded dual variables and power accumulator, and distributes it along the link; S2. When the boundary deduction module of the target node captures the issued dynamic token, it collects the local ambient temperature and AC voltage, and respectively deduces the thermal and grid critical safety power and takes the intersection to establish the local feasible region. S3. The target node's drive execution module extracts the dual variable from the dynamic token, projects the parsed nominal expected power into the established local feasible region in combination with the dual variable to obtain the local allocated power, and accumulates the local allocated power into the power accumulator to update the dynamic token and pass it to the next node. S4. When the updated dynamic token is transferred to the tail node, the tail node calculates the power surplus and deficit of the entire network based on the power accumulator that has been accumulated, updates the dual variable and determines whether convergence has occurred. S5. If convergence is determined, each drive execution module will convert the projected local allocated power into a voltage value and write it into a register to execute the physical drive; if convergence is determined not to be achieved, the dynamic token containing the new dual variable will be sent back to the first node to re-trigger the deduction process.

2. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S1, the cascaded parsing module of the first node parses the configuration packet to obtain the nominal expected power, including: The cascading parsing module of the first node receives communication data streams through a network interface; The communication data stream is unpacked and subjected to cyclic redundancy check to extract the global expected total power; Send topology probe frames to obtain the total number of online LED driver power supplies; The nominal expected power is obtained by dividing the global expected total power by the total number of LED driver power supplies.

3. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S1, generating a dynamic token with embedded dual variables and a power accumulator and distributing it along the link includes: Allocate contiguous blocks of data in the microcontroller's memory; The address spaces corresponding to the dual variables and the power accumulator in the data block are cleared and initialized; Add a frame header identifier and the target routing address to the data block and encapsulate it into the dynamic token; The dynamic token is pushed into the send buffer of the serial communication interface for distribution.

4. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S2, the separate deduction of thermal and grid critical safety power includes: Read the analog-to-digital conversion sample value of the external thermistor; The maximum driving current allowed to pass under the current heat dissipation boundary is calculated by calling the preset steady-state thermal resistance matrix, and the maximum driving current is mapped to the thermal critical safe power. The transient sag slope is calculated by acquiring the zero-crossing time and peak value data of the AC voltage. By combining the impedance factor of the power supply phase and the transient drop slope, the critical safe power of the power grid to prevent AC bus undervoltage can be deduced.

5. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S2, the step of taking the intersection to establish the local feasible region includes: A numerical comparison calculation is performed between the derived thermal critical safety power and the power grid critical safety power; Choose the smaller value as the upper limit of the absolute safe power; The minimum dimming power required to maintain continuous oscillation of the internal switching power supply chip is taken as the absolute safe power lower bound. The lower bound of the absolute safe power and the upper bound of the absolute safe power are combined to construct a continuous mathematical closed interval, forming the local feasible region.

6. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S3, the step of projecting the analytically obtained nominal expected power, combined with the dual variable, onto the established local feasible region to obtain the local allocated power includes: The theoretical compensatory power is obtained by calculating the difference between the nominal expected power and the dual variable; Perform a one-dimensional Euclidean projection operation to compare the theoretical compensation power with the upper bound and the lower bound of the absolute safe power of the local feasible region. The theoretical compensatory power is truncated under the constraint of not exceeding the upper and lower bounds of the absolute safe power, and the local allocated power is output.

7. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S3, accumulating the locally allocated power to the power accumulator to update the dynamic token and passing it to the next node includes: Extract the old power accumulation value written by the previous node from the payload segment of the captured dynamic token; The calculated local power allocation is added to the old power accumulation value to obtain a new power accumulation value; The new power accumulation value is overwritten into the payload segment of the dynamic token; The updated dynamic token is sent to the next node in the physical topology via the transmit pin of the serial communication interface.

8. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S4, the tail node calculates the power surplus / deficit of the entire network based on the power accumulator after accumulation, and updates the dual variables including: Extract the preset subgradient iteration step size parameter; The final new power accumulation value in the dynamic token flowing to the tail node is subtracted from the global expected total power to obtain the overall network power surplus or deficit. The step bias is obtained by multiplying the total network power surplus / deficit with the subgradient iteration step size parameter. The update operation is completed by adding the step bias to the current dual variable.

9. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S4, determining whether convergence has occurred includes: Calculate the absolute difference between the updated dual variable and the original dual variable; Extract the convergence threshold parameters that are pre-stored in non-volatile memory; Compare the absolute difference with the convergence threshold parameter. If the absolute difference is less than or equal to the convergence threshold parameter, output a Boolean truth value indicating convergence; If the absolute difference is greater than the convergence threshold parameter, output a Boolean false value indicating that convergence has not been achieved.

10. The remote parameter configuration method for an LED driver power supply according to claim 1, characterized in that, In S5, each drive execution module converts the projected local allocated power into a voltage value and writes it into a register to execute the physical drive, including: Receive the convergence lock command broadcast by the tail node; The microcontroller's internal digital-to-analog conversion mapping table is invoked to convert the locally allocated power retained in the last iteration into the corresponding target reference voltage digital value; Write the target reference voltage digital value into the control register of the digital-to-analog converter; The analog voltage is output to the feedback pin of the pulse width modulation controller to adjust the duty cycle of the power switch.

Images