Full-spectrum LED street lamp system based on remote control and energy-saving analysis

Abstract

The invention discloses a full-spectrum LED street lamp system based on remote control and energy-saving analysis, and relates to the technical field of LED illumination control, and the system comprises a multi-source data sensing module which collects and standardizes data in real time, and a dual-mode cooperative communication module which efficiently transmits data and instructions; the PINN constraint decision module generates a precise control scheme and performs multi-dimensional error analysis and optimization; the single lamp execution module realizes smooth dimming and effect acquisition; the data feedback iterative optimization module dynamically adjusts model parameters and constraint thresholds through deviation analysis and incremental learning to form a closed-loop optimization system; according to the method, multi-source data perception, dual-mode communication and PINN constraint decision are fused, the LED street lamp dynamically adapts to road, traffic and environmental conditions, the energy consumption is reduced while the illumination quality is guaranteed, smooth brightness adjustment is ensured through feedback iteration optimization model parameters and a PWM dimming technology, and intelligent and efficient development of a street lamp system is integrally promoted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of LED lighting control, in particular to a full-spectrum LED street lamp system based on remote control and energy-saving analysis. BACKGROUND

[0002] With the acceleration of urbanization and the improvement of intelligent transportation system, as an important part of urban infrastructure, the functional requirements of road lighting have been expanded from simple night lighting to energy saving, intelligentization and self-adaptation. LED light source, with the advantages of high efficiency, long service life and environmental protection, gradually replaces traditional light source and becomes the mainstream choice of road lighting. The development of full-spectrum LED technology further improves the lighting comfort and color rendering, meeting the visual needs in different scenarios. Meanwhile, the integration of remote control, Internet of Things and artificial intelligence provides technical support for the intelligent upgrading of road lighting system, making it possible to dynamically adjust lighting parameters according to environmental changes, traffic flow and other factors, ensuring road safety and meeting the development concept of energy saving and emission reduction, and promoting the transformation of road lighting system to fine and efficient management.

[0003] The traditional road lighting system has many limitations in practical application and cannot meet the diversified needs of modern urban lighting. In terms of control mode, fixed brightness mode or simple time sequence control is often used, which lacks dynamic response ability to real-time environmental conditions and traffic flow changes, resulting in mismatch between lighting effect and actual demand. Either excessive lighting causes energy waste or insufficient lighting affects traffic safety. In terms of data transmission, single communication mode is easily disturbed by environment, has transmission delay and poor stability, and cannot guarantee efficient transmission of remote control instructions and accuracy of data interaction. In terms of decision logic, there is a lack of deep integration of physical laws and scene constraints, and the scientificity and adaptability of control scheme are insufficient. Moreover, there is no perfect feedback optimization mechanism, and the system cannot continuously adjust parameters according to actual effect during operation, resulting in unstable lighting quality. In addition, the operation and maintenance of traditional system rely on manual inspection, which is not only low in efficiency and high in cost, but also difficult to find equipment failure and performance degradation problems in time, affecting the long-term stable operation of the system. SUMMARY

[0004] The present application aims to overcome the shortcomings of the prior art and provides a full-spectrum LED street lamp system based on remote control and energy-saving analysis. The system can collect and process scene and equipment data in real time through multi-source data sensing and standardization modules, upload data and receive control instructions through a dual-mode collaborative communication module, generate a control scheme that meets physical laws through a PINN constraint decision module, realize smooth dimming and effect collection through a single lamp execution module, and dynamically adjust model parameters and constraint conditions through a data feedback iterative optimization module through a feedback mechanism to form a closed-loop optimization system, effectively improving the energy efficiency and management level of street lamps.

[0005] The application provides the following technical scheme to solve the above technical problems: a full-spectrum LED street lamp system based on remote control and energy-saving analysis, which comprises: A multi-source data sensing and standardization module: scene sensing data, device state data and operating parameter data are collected in real time by a sensor group, abnormal data is removed by using a 3 sigma rule, and after min-max standardization processing, the data is integrated into a structured data package according to a preset structure; A dual-mode cooperative communication module: receives the structured data package, uploads the scene and device state data by using the URLLC slice technology of a 5G / 6G communication link, and transmits the operating parameter data by combining PLC technology and OFDM technology; according to a preset priority, a link forwards a remote control instruction; A PINN constraint decision module: structured data is input into a physical information neural network model, a control scheme is initially generated in combination with a preset constraint condition, a physical quantity is calculated and predicted by using an embedded physical equation, an optimal solution is selected by multi-dimensional error analysis, when the error exceeds an allowable range, the model parameters are iteratively optimized by using an Adam optimizer, and finally a standardized control instruction set conforming to physical laws and actual requirements is output; A single lamp execution and effect collection module: receives the control instruction set, adjusts the parameters of each primary color LED by using an LED drive circuit, and realizes smooth dimming by using a PWM dimming technology; lighting effects, running and device state feedback data are synchronously collected and arranged into an execution effect data package in a unified format; A data feedback iterative optimization module: receives the execution effect data package and returns it to the control center, performs deviation analysis on the feedback data and predicted data by using a feedback weighted deviation optimization algorithm, iteratively optimizes the model parameters by using an incremental learning method, and dynamically adjusts the threshold of the preset constraint condition.

[0006] Further, in the multi-source data sensing and standardization module, the sensor group is composed of a scene sensing sensor sub-group, a device state monitoring sensor sub-group and an operating parameter acquisition sensor sub-group; the scene sensing sensor sub-group comprises an ambient illuminance sensor, an integrated weather sensor, a roadside geomagnetic sensor, a video recognition camera and a road type identification sensor, which respectively collect ambient illuminance, weather conditions and visibility, vehicle passing frequency, vehicle quantity and speed, and road type identification information; the device state monitoring sensor sub-group comprises a brightness feedback module, a drive circuit monitoring unit and a running time counter, which respectively collect real-time brightness of the street lamp, drive circuit temperature and current-voltage fluctuation, cumulative running time and brightness attenuation related data of the street lamp; the operating parameter acquisition sensor sub-group comprises a voltage sensor, a current sensor, a power factor acquisition module and an energy consumption monitoring unit, which respectively collect real-time voltage, real-time current, power factor and real-time energy consumption data in the working process of the street lamp.

[0007] Further, in the multi-source data perception and standardization module, the structured data packet adopts a three-layer structure of header + body + check code: the header contains a timestamp accurate to milliseconds, a 16-bit binary encoded unique device ID of a single lamp, and a data type identifier; the body is JSON format data containing road type code, traffic flow period code, illumination value, cumulative running time, brightness attenuation rate, voltage, and current information; the check code is generated by a CRC-32 algorithm for verifying data transmission integrity, and the specific execution process of the CRC-32 algorithm is as follows: initialize a 32-bit register value to 0xFFFFFFFF; take each byte of the header and body of the structured data packet in turn, and perform an XOR operation between the byte and the lower 8 bits of the 32-bit register; right shift the XOR operation result by 1 bit, if the removed bit is 1, perform an XOR operation between the register value and the CRC-32 standard polynomial 0x04C11DB7, if the removed bit is 0, keep the register value unchanged, repeat the right shift and XOR operation 8 times to complete the processing of the current byte; after all bytes are processed, perform an XOR operation between the 32-bit register value and 0xFFFFFFFF, and the result obtained is the final CRC-32 check code.

[0008] Further, in the dual-mode cooperative communication module, the 5G / 6G communication link adopts URLLC slicing technology, the slicing bandwidth allocation range is 100Mbps to 1Gbps, the peak transmission rate is ≥10Gbps, and the end-to-end delay is ≤10ms; the scene and device state data upload adopts a fragmentation transmission mechanism, the single packet size is limited to 1KB to 10KB, the retransmission is triggered when a fragment is lost, the retransmission number is ≤3 times, and the retransmission timeout time is 50ms.

[0009] Further, in the dual-mode cooperative communication module, the PLC communication link adopts OFDM modulation and demodulation technology, the subcarrier spacing is 1kHz to 20kHz, the carrier frequency range is 2MHz to 30MHz, and the communication rate is 1Mbps to 10Mbps; the operation parameter data transmission adopts a frame structure + address coding mode, each frame contains a target street lamp device address, data length, data content, and a check bit, and only the street lamp terminal with address matching receives and analyzes the data.

[0010] Further, in the dual-mode cooperative communication module, the preset priority is divided according to the influence degree of instructions on system operation and the response timeliness requirement, and is divided into four levels from high to low: the first level is an emergency fault shutdown instruction; the second level is a spectrum adjustment instruction; the third level is a regular dimming instruction; and the fourth level is a state query instruction; the first level instruction is forwarded through the 5G / 6G link, and the second to fourth level instructions are preferentially transmitted through the PLC link.

[0011] Further, in the PINN constraint decision module, the physical information neural network model comprises an input layer, a hidden layer, an output layer and a physical constraint layer: the input layer has 18 neurons corresponding to the scene, device and running characteristics; the hidden layer has 3-5 layers, each layer has 64-256 neurons, and the activation function uses ReLU function; the output layer has 6 neurons corresponding to the currents of four primary color LEDs, the dimming ratio and the switch state.

[0012] Further, in the PINN constraint decision module, the specific process of initially generating a control scheme by inputting structured data into the physical information neural network model and combining the preset constraint conditions is as follows: first, the received structured data is feature-split to extract the road type code, traffic flow period code, illumination value and weather condition code in the scene perception data, the cumulative running time, brightness attenuation rate and driving circuit temperature in the device state data, and the real-time voltage, current, power factor and real-time energy consumption characteristic term in the running parameter data, and each characteristic term is mapped to an input layer neuron of the physical information neural network model to complete the input initialization of the feature vector; the preset constraint conditions are divided into three constraint groups according to the road type, the first group is the main road constraint group: the lighting uniformity threshold is greater than or equal to 0.5, the energy consumption threshold in the peak period is less than or equal to 150 W, the energy consumption threshold in the trough period is less than or equal to 80 W, the color temperature threshold range is 2700K to 6500K, and the color rendering index threshold is greater than or equal to 85; the second group is the secondary road constraint group: the lighting uniformity threshold is greater than or equal to 0.45, the energy consumption threshold in the peak period is less than or equal to 120 W, the energy consumption threshold in the trough period is less than or equal to 70 W, the color temperature threshold range is 2700K to 6500K, and the color rendering index threshold is greater than or equal to 80; the third group is the branch road constraint group: the lighting uniformity threshold is greater than or equal to 0.4, the energy consumption threshold in the peak period is less than or equal to 100 W, the energy consumption threshold in the trough period is less than or equal to 60 W, the color temperature threshold range is 2700K to 6500K, and the color rendering index threshold is greater than or equal to 75; the model matches the corresponding constraint group according to the road type code in the structured data, performs nonlinear feature conversion through the feature extraction hidden layer based on the input feature vector, preliminarily filters the converted feature vector in combination with the matched constraint group, eliminates the feature combinations exceeding the constraint threshold, and then calculates the initial predicted physical quantity through the illumination superposition law and the Lambert cosine law embedded in the physical constraint fusion layer to finally generate a preliminary control scheme containing the driving currents of each primary color LED, the dimming ratio and the switch state.

[0013] Further, in the PINN constraint decision module, the specific process of calculating the predicted physical quantity by using the built-in physical equation is as follows: first, the core parameters output by the model feature extraction layer are extracted, including the luminous flux of the primary color LED , the street lamp installation height, the primary color LED light emitting angle, the driving current , and the light source installation elevation angle; the target point on the road surface is calculated through the illumination superposition law The initial illuminance value, corresponding to the mathematical expression is: ,in, Target locations on the road surface initial illuminance, These are the pre-set target monitoring points on the road surface. This refers to the number of primary color LEDs included in a full-spectrum LED street light. For the first Normal and position of each primary color LED light source The angle between directions, For the LED to the point Spatial distance, No. The luminous flux of a primary color LED The sphericity coefficient; the initial illuminance is corrected using Lambert's cosine law to obtain the point position. The corrected predicted illuminance is expressed as: ,in, Target location Predicted illuminance corrected by Lambert's cosine law Install elevation angle and position of light source The composite angle of azimuth; based on multiple road surface target points. Illumination uniformity can be predicted using the illumination uniformity calculation formula, which is: , Predicting lighting uniformity, It is the minimum corrected illuminance among multiple road surface target points. It is the arithmetic mean of the corrected illuminance at multiple road surface target points; combined with the human visual function, the tristimulus values ​​are obtained by weighting the spectra of each primary color LED. , , Substitute the values ​​into the color temperature calculation formula to predict the color temperature. The color temperature calculation formula is: , It predicts color temperature. It is a color temperature conversion function, and the predicted color rendering index is obtained based on the comparison results of the tristimulus values ​​and the CIE standard light source; simultaneously, it is based on the driving current of each primary color LED. and corresponding operating voltage Calculate the power of a single primary color LED The formula is: The predicted energy consumption value is obtained by summing the results. The formula is: , It is the predicted total energy consumption of full-spectrum LED streetlights, which ultimately forms a set of predicted physical quantities including the predicted illuminance, predicted lighting uniformity, predicted color temperature, predicted color rendering index and predicted energy consumption value after correction at each point.

[0014] Further, in the PINN constraint decision module, the multi-dimensional error includes data fitting error, physical law deviation, constraint satisfaction deviation, and dynamic adaptation deviation. The data fitting error is the arithmetic mean of the absolute deviation of each control parameter in the preliminary control scheme from the corresponding parameters of the historical optimal sample set of the same scene, and the threshold value is set to 0.04. The physical law deviation is the maximum of the relative deviations of the predicted physical quantity from the theoretical calculation value of the built-in physical equation, and the threshold value is set to 0.03. The constraint satisfaction deviation is the normalized deviation maximum when the predicted physical quantity exceeds the preset constraint condition threshold value, and the deviation value is 0 when it does not exceed the threshold value. The threshold value is set to 0.02. The dynamic adaptation deviation is the adaptation score deviation value calculated based on the scene real-time change rate and the historical same period change rule. The adaptation score range is 0-1, and the deviation value is obtained by subtracting the adaptation score by 1. The threshold value is set to 0.05.

[0015] Further, in the PINN constraint decision module, the specific process of filtering the optimal solution through multi-dimensional error analysis is as follows: the threshold values of each error and the total error weight are set in advance: the data fitting error threshold is 0.04, the physical law deviation threshold is 0.03, the constraint satisfaction deviation threshold is 0.02, and the dynamic adaptation deviation threshold is 0.05. The data fitting error weight is 0.2, the physical law deviation weight is 0.3, the constraint satisfaction deviation weight is 0.3, and the dynamic adaptation deviation weight is 0.2. First, the specific values of the four single errors are calculated, and it is judged whether each error is within the corresponding threshold value. Then, the multi-dimensional total error is calculated according to total error = data fitting error × 0.2 + physical law deviation × 0.3 + constraint satisfaction deviation × 0.3 + dynamic adaptation deviation × 0.2, and the total error threshold is set to 0.05. If all four single errors do not exceed the corresponding threshold value, and the multi-dimensional total error is ≤0.05, it is determined that the preliminary control scheme is the optimal solution. If any single error exceeds the corresponding threshold value, or the multi-dimensional total error is >0.05, it is determined that the preliminary control scheme does not meet the standard, and the subsequent model parameter iteration optimization process is triggered.

[0016] Further, the single lamp performs with the effect collection module, the LED drive circuit includes independent constant current drive unit and parameter collaborative control unit, each primary color LED corresponds to exclusive constant current drive unit, after the parameter collaborative control unit receives the drive current, voltage parameter in the standardized control instruction set, the accurate control signal is output to each constant current drive unit, the working current and voltage of each primary color LED are adjusted, the independent regulation and control of each primary color luminous flux are realized;PWM dimming technology generates dimming signal through the high-frequency signal generator built-in in the drive circuit, the dimming frequency is set to 30 kHz, the parameter collaborative control unit adjusts the duty cycle of the dimming signal in the range of 0% to 100% linearly according to the dimming proportion in the control instruction, the adjustment step of the duty cycle is 0.1%, and the duty cycle change rate is set through the gradual change control logic, so that the transition time between adjacent duty cycles is not less than 0.5s, through the collaborative output of the parameter regulation and control of each primary color LED and the PWM dimming signal, the smooth change of the street lamp brightness is realized.

[0017] Further, in the data feedback iterative optimization module, the mathematical expression of the feedback weighted deviation optimization algorithm is: ; wherein: is the comprehensive deviation value, is the total number of data items participating in deviation analysis, is the actual value of the th feedback data in the execution effect data packet, is the th predicted data output by the PINN constraint decision module, is the reference benchmark value corresponding to the th data, is the minimum value, is the importance weight of the th data, and , is the equipment aging correction factor, is the scene dynamic coefficient, is the timeliness decay coefficient of the th data.

[0018] Furthermore, in the data feedback iterative optimization module, the specific process of iteratively optimizing model parameters and dynamically adjusting preset constraint thresholds through incremental learning is as follows: First, the received execution effect data packet is preprocessed to extract lighting effect feedback data, operation status feedback data, and equipment status feedback data, which are then matched with the corresponding predicted physical quantities. Feedback data deemed valid in the deviation analysis is retained, while abnormal data is removed. When using incremental learning, a fixed sample window is set, maintaining 50-100 sets of valid data within the window. When new valid feedback data is added, the earliest equivalent amount of old data within the window is removed in chronological order to form an incremental learning sample set. During the iterative optimization of model parameters stage, the underlying features of the physical information neural network model are frozen. Layer parameters are selected, and only the weights and bias parameters of the top fully connected layer and the physical constraint fusion layer are fine-tuned. The Adam optimizer is used, with a learning rate of 0.001 and 50 iterations. The model output error is calculated after each iteration, and parameter optimization stops when the error converges. When dynamically adjusting the preset constraint thresholds, the adjustment operation is performed according to road type. For each type of road, the average deviation between 10 consecutive sets of valid feedback data and predicted data is calculated. If the average deviation is ≤0.02, the lighting uniformity constraint threshold for that type of road is increased by 0.02 and the energy consumption constraint threshold is decreased by 5W, with the adjustment step not exceeding 10% of the initial threshold. If the average deviation is ≥0.06, the lighting uniformity constraint threshold is decreased by 0.02 and the energy consumption constraint threshold is increased by 5W.

[0019] Compared with existing technologies, this full-spectrum LED street light system based on remote control and energy-saving analysis has the following advantages: I. This invention integrates multi-dimensional information on scenarios, equipment, and operation through multi-source data perception and standardized processing. It combines dual-mode collaborative communication technology to achieve efficient data transmission and hierarchical command forwarding. Relying on the PINN constraint decision module, it incorporates physical laws and scenario constraints to generate precise control schemes. Through multi-dimensional error analysis, it selects the optimal solution. Combined with the refined control and effect feedback of the single-lamp execution link, it constructs a closed-loop optimization system. This system not only achieves dynamic adaptation of lighting parameters to road type, traffic flow, and environmental conditions to ensure lighting quality in different scenarios, but also reduces energy consumption through real-time energy consumption monitoring and intelligent dimming strategies. At the same time, the remote control function reduces manual maintenance costs and improves the convenience and reliability of system operation.

[0020] Second, this invention ensures that the control scheme conforms to physical laws and actual needs through the deep integration of physical information neural networks and constraints. It continuously improves model parameters and constraint thresholds through a data feedback iterative optimization mechanism, enhancing the system's dynamic adaptability. The collaborative application of dual-mode communication balances transmission rate and stability. The hierarchical instruction forwarding mechanism ensures rapid response to key instructions. Comprehensive perception from multiple sources and standardized data processing provide accurate data support for decision-making. PWM dimming technology enables smooth brightness adjustment, avoiding the adverse effects of sudden changes in illumination. The overall architecture takes into account lighting effect, energy saving efficiency, and operation and maintenance efficiency, promoting the development of LED street light systems towards intelligence, high efficiency, and self-adaptability, and is suitable for various road lighting scenarios.

[0021] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0023] Figure 1 A flowchart of a full-spectrum LED street light system based on remote control and energy-saving analysis; Figure 2 This is a schematic diagram of data transmission between modules in a full-spectrum LED street light system based on remote control and energy-saving analysis. Figure 3 This is a schematic diagram of data transmission between PINN constraint decision modules. Detailed Implementation

[0024] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0025] Example 1: Example of operation during peak hours on urban main roads.

[0026] The sensor group fully commenced data acquisition. The ambient illuminance sensor in the scene perception sensor subgroup accurately captured current ambient light conditions; the integrated meteorological sensor monitored weather and visibility in real time to adapt to different lighting needs; the roadside geomagnetic sensor efficiently monitored vehicle passing frequency; the video recognition camera accurately counted the number and speed of vehicles; and the road type identification sensor identified the road segment as an urban arterial road and obtained its code. The equipment status monitoring sensor subgroup monitored street light illumination in real time through a brightness feedback module; the drive circuit monitoring unit meticulously recorded circuit operation status to prevent malfunctions; and the runtime counter accurately tracked street light usage to provide a basis for brightness decay assessment. The operating parameter acquisition sensor subgroup comprehensively collected data such as voltage and current to ensure accurate energy consumption monitoring. After acquisition, the 3σ criterion was used to effectively eliminate outlier data to ensure data reliability. Min-max standardization ensured the data had a unified comparability standard, and the data was integrated into a structured data packet according to a three-layer structure of header + body + checksum. The timestamp, device ID, and data type identifier in the header ensured data traceability; the JSON format body centrally presented key information for subsequent processing; and the checksum generated by the CRC-32 algorithm strictly ensured that no missing or erroneous data occurred during data transmission. Figure 1 As shown.

[0027] After receiving structured data packets, the dual-mode collaborative communication module efficiently allocates data transmission tasks according to preset rules. Scene and device status data are uploaded using URLLC slicing technology on 5G / 6G communication links. This technology's high bandwidth and low latency ensure fast and accurate data transmission. The slicing transmission mechanism and retransmission rules further guarantee data integrity, preventing data loss from affecting subsequent decisions. Operating parameter data is transmitted using a combination of PLC and OFDM technology. The OFDM modulation and demodulation technology of the PLC communication link adapts to complex transmission environments. Frame structure and address encoding ensure that data is only received by the target streetlight terminal, improving the security and targeting of data transmission. Command transmission allocates links according to priority. Level 1 emergency fault shutdown commands are quickly responded to via the 5G / 6G link, while level 2 to 4 commands are preferentially transmitted via the PLC link, rationally allocating communication resources to improve overall transmission efficiency.

[0028] The PINN constraint decision module inputs structured data packets into the physical information neural network model. Each layer of the model performs its specific function: the input layer neurons precisely correspond to features to ensure comprehensive information input; the hidden layer's nonlinear feature transformation capability deeply mines data correlations; the output layer explicitly outputs control-related parameters; and the physical constraint layer embeds relevant laws to ensure the control scheme conforms to objective laws. After feature decomposition of the structured data, each feature term is mapped one-to-one with the input layer neurons for precise initialization. Main road constraint groups are matched according to road type codes. The indicators of this constraint group meet the lighting requirements of the main road during peak hours while effectively controlling energy consumption. The core parameters output by the model's feature extraction layer are first extracted, including... The luminous flux of each primary color LED Street light installation height, light emission angle of each primary color LED, and drive current. 1. Light source installation elevation angle; 2. Calculation of target points on the road surface using the law of superposition of illuminance. The initial illuminance value, corresponding to the mathematical expression is: ,in, Target locations on the road surface initial illuminance, These are the pre-set target monitoring points on the road surface. This refers to the number of primary color LEDs included in a full-spectrum LED street light. For the first Normal and position of each primary color LED light source The angle between directions, For the LED to the point Spatial distance, No. The luminous flux of a primary color LED The sphericity coefficient; the initial illuminance is corrected using Lambert's cosine law to obtain the point position. The corrected predicted illuminance is expressed as: ,in, Target location Predicted illuminance corrected by Lambert's cosine law Install elevation angle and position of light source The composite angle of azimuth; based on multiple road surface target points. Illumination uniformity can be predicted using the illumination uniformity calculation formula, which is: , Predicting lighting uniformity, It is the minimum corrected illuminance among multiple road surface target points. It is the arithmetic mean of the corrected illuminance at multiple road surface target points; combined with the human visual function, the tristimulus values ​​are obtained by weighting the spectra of each primary color LED. , , Substitute the values ​​into the color temperature calculation formula to predict the color temperature. The color temperature calculation formula is: , It predicts color temperature. It is a color temperature conversion function, and the predicted color rendering index is obtained based on the comparison results of the tristimulus values ​​and the CIE standard light source; simultaneously, it is based on the driving current of each primary color LED. and corresponding operating voltage Calculate the power of a single primary color LED The formula is: The predicted energy consumption value is obtained by summing the results. The formula is: , This is the predicted total energy consumption of full-spectrum LED streetlights, ultimately forming a set of predicted physical quantities including predicted illuminance, predicted lighting uniformity, predicted color temperature, predicted color rendering index, and predicted energy consumption values ​​after corrections at each location. Subsequently, a multi-dimensional error analysis is used to comprehensively evaluate the feasibility of the scheme. Reasonable settings of various error indicators ensure the accuracy and adaptability of the control scheme. If the error exceeds the allowable range, the Adam optimizer iteratively optimizes the model parameters. The final standardized control instruction set output conforms to physical laws and meets the actual lighting and energy-saving needs of main roads during peak hours. Figure 3 As shown.

[0029] After receiving the control command set, the single-lamp execution and effect acquisition module works closely with the independent constant current drive unit and parameter coordination control unit of the LED driver circuit to precisely adjust the working current and voltage of each primary color LED, achieving independent control of the luminous flux of each primary color and ensuring that the spectrum and brightness meet the requirements. PWM dimming technology uses a dimming signal of a specific frequency to linearly adjust the duty cycle according to the dimming ratio in the control command, and the gradual control logic ensures smooth changes in street light brightness, avoiding sudden changes in light that could affect driving safety. Simultaneously, lighting effect, operation, and equipment status feedback data are collected and organized into an execution effect data package in a unified format, providing comprehensive and standardized actual data support for subsequent deviation analysis and model optimization.

[0030] The data feedback iterative optimization module receives the execution effect data packet and sends it back to the control center. The feedback weighted deviation optimization algorithm comprehensively analyzes the deviation between the feedback data and the predicted data. The mathematical expression of the feedback weighted deviation optimization algorithm is: ;in: This is the overall deviation value. The total number of data items participating in the deviation analysis. For the execution effect data packet, the first The actual value of the feedback data. The first output of the PINN constraint decision module Item forecast data, For the first The reference baseline value corresponding to each data item It is the minimum value. For the first The importance weight of each data item, and , For equipment aging correction factors, For scene dynamic coefficients, multiple dimensions are considered, including data importance, equipment status, and scene changes, to ensure accurate and reliable deviation analysis results. Incremental learning dynamically updates the learning sample set through a fixed sample window, ensuring model optimization is based on the latest valid data. Parameters of the underlying feature extraction layer are frozen, with only relevant layer parameters fine-tuned, improving model adaptability while avoiding overtraining. Preset constraint thresholds are dynamically adjusted according to road type, and lighting uniformity and energy consumption thresholds are reasonably adjusted based on the mean deviation, ensuring constraints continuously adapt to actual operating conditions and constantly improving the system's lighting effect and energy efficiency.

[0031] In summary, in this embodiment, the full-spectrum LED street light system on the urban main road operates in an orderly manner during peak hours. The multi-source data sensing and standardization module comprehensively collects and processes various types of data, ensuring reliable and usable data; the dual-mode collaborative communication module rationally allocates communication resources to achieve fast and secure data transmission; the PINN constraint decision module generates the optimal control scheme based on accurate data and scientific constraints, meeting the lighting and energy-saving needs during peak hours; the single-lamp execution and effect acquisition module accurately executes instructions and feeds back actual data; and the data feedback iterative optimization module continuously optimizes the model and constraints. The entire process is tightly integrated, ensuring sufficient lighting and driving safety on the main road during peak hours while effectively controlling energy consumption through scientific regulation, fully leveraging the system's remote control and energy-saving advantages.

[0032] Example 2: Example of operation of urban side roads during off-peak hours at night.

[0033] The sensor group comprehensively collects data. The scene perception sensor subgroup includes an ambient light sensor capturing low-light conditions during nighttime off-peak hours, an integrated meteorological sensor acquiring current weather and visibility information, a roadside geomagnetic sensor monitoring low vehicle traffic frequency, a video recognition camera identifying low vehicle numbers and slow speeds, and a road type identification sensor confirming the road segment as an urban branch road and obtaining its code. The equipment status monitoring sensor subgroup comprehensively collects data such as real-time streetlight brightness, drive circuit operating status, and cumulative operating time, providing a basis for equipment status assessment. The operating parameter acquisition sensor subgroup accurately collects voltage, current, power factor, and real-time energy consumption data, ensuring accurate energy consumption monitoring. After collection, the 3σ criterion effectively removes outlier data, min-max standardization unifies the data format, and the data is integrated into a structured data packet with a three-layer structure of header + body + checksum. The header information ensures data traceability, the body data in JSON format centrally presents key parameters, and the checksum generated by the CRC-32 algorithm ensures the integrity and error-free transmission of data, providing a reliable foundation for subsequent data processing and decision-making. Figure 2 As shown.

[0034] After receiving structured data packets, the dual-mode collaborative communication module scientifically allocates transmission tasks. Scene and device status data are uploaded via URLLC slicing technology on 5G / 6G communication links. The high transmission rate and low latency of this technology ensure rapid data upload, while the slicing transmission mechanism and retransmission rules ensure no data loss. Operating parameter data is transmitted via a combination of PLC and OFDM technology. The relevant parameter settings of the PLC communication link are adapted to the transmission environment of the branch, and the frame structure + address encoding method ensures accurate data transmission to the target terminal, avoiding data transmission chaos. Command transmission allocates links according to priority. First-level commands respond quickly, while second to fourth-level commands prioritize the use of the PLC link, making reasonable use of communication resources to ensure efficient and stable data transmission and meet the operational needs of branch streetlights during off-peak hours.

[0035] After structured data packets are input into the physical information neural network model, the various layers of the model work collaboratively. The input layer comprehensively receives feature data, the hidden layer deeply mines data correlations, the output layer specifies control parameters, and the physical constraint layer ensures that the control scheme conforms to objective laws. After feature decomposition, each feature item is precisely mapped to the input layer neurons. Based on road type coding, a branch road constraint group is matched. The indicators of this constraint group not only meet the basic lighting needs of the branch road during off-peak hours at night but also maximize energy saving. The model generates a preliminary control scheme through a series of processes. Multi-dimensional error analysis comprehensively evaluates the scheme from aspects such as data fitting, physical laws, constraint satisfaction, and dynamic adaptation. The reasonable setting of various error thresholds and weights ensures the reliability of the scheme. If the scheme does not meet the standards, iterative optimization is performed through the Adam optimizer. The final output standardized control instruction set is adapted to the low traffic flow and energy saving needs of the branch road during off-peak hours at night.

[0036] After receiving the control command set, the single-lamp execution and effect acquisition module precisely adjusts the parameters of each primary color LED to ensure that the lighting effect meets the nighttime usage requirements of the branch. PWM dimming technology adjusts the duty cycle according to the specified frequency and step size, and the gradual control logic achieves smooth brightness changes, avoiding sudden changes in light that may affect pedestrians or vehicles. The synchronously acquired lighting effect, operation, and equipment status feedback data are organized into an execution effect data package in a unified format, comprehensively recording the actual operation. This provides accurate and standardized data support for subsequent deviation analysis and model optimization, contributing to the continuous improvement of the system.

[0037] The data feedback iterative optimization module receives execution effect data packets and transmits them back to the control center. The feedback weighted deviation optimization algorithm comprehensively analyzes the deviation between the feedback data and the predicted data from multiple factors, ensuring that the analysis results are comprehensive and accurate. Incremental learning dynamically updates the sample set, allowing model optimization to be based on the latest effective data and enabling targeted fine-tuning of model parameters to improve the model's adaptability to nighttime off-peak hours. Dynamically adjusting preset constraint thresholds based on the mean deviation better adapts the constraints to the actual nighttime operating conditions of the branch roads, further optimizing energy-saving effects while ensuring basic lighting, and promoting the long-term stable and efficient operation of the system.

[0038] In summary, this embodiment demonstrates the efficient operation of the full-spectrum LED streetlight system on urban side roads during off-peak hours at night. The multi-source data sensing and standardization module accurately captures the characteristics of low traffic flow and low illumination on side roads, generating reliable structured data. The dual-mode collaborative communication module adapts to the side road transmission environment, achieving efficient data transmission. The PINN constraint decision module matches the side road constraint group, generating a control scheme that balances basic lighting and energy saving. The single-lamp execution and effect acquisition module smoothly executes commands and provides data feedback. The data feedback iterative optimization module continuously optimizes system parameters. The entire process aligns with the actual needs of side roads during off-peak hours at night, maximizing energy savings while ensuring basic lighting safety for pedestrians and vehicles, showcasing the system's flexibility and practicality.

[0039] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A full spectrum LED street light system based on remote control and energy saving analysis, characterized in that, The system comprises: A multi-source data perception and standardization module: real-time collection of scene perception data, equipment state data and operating parameter data through a sensor group, elimination of abnormal data using the 3σ criterion, integration into a structured data package after min-max standardization, and forwarding of remote control instructions according to a preset priority selection link; A dual-mode collaborative communication module: receiving the structured data package, uploading the scene and equipment state data by means of the URLLC slice technology of the 5G / 6G communication link, and transmitting the operating parameter data by combining the PLC technology and the OFDM technology; A PINN constraint decision module: inputting the structured data into a physical information neural network model, generating a control scheme in combination with preset constraint conditions, calculating predicted physical quantities using built-in physical equations, selecting the optimal solution through multi-dimensional error analysis, and outputting a standardized control instruction set that meets the physical laws and actual requirements when the error exceeds the allowable range, and iteratively optimizing the model parameters using the Adam optimizer; A single lamp execution and effect collection module: receiving the control instruction set, adjusting the parameters of each primary color LED through the LED drive circuit, and achieving smooth dimming using the PWM dimming technology; synchronously collecting lighting effects, running and equipment state feedback data, and arranging them into an execution effect data package in a unified format; A data feedback iterative optimization module: receiving the execution effect data package and returning it to the control center, performing deviation analysis on the feedback data and predicted data through a feedback weighted deviation optimization algorithm, iteratively optimizing the model parameters through incremental learning, and dynamically adjusting the threshold of the preset constraint conditions.

2. The full spectrum LED street light system based on remote control and energy saving analysis according to claim 1, characterized in that, In the multi-source data perception and standardization module, the sensor group is composed of a scene perception sensor subgroup, an equipment state monitoring sensor subgroup and an operating parameter acquisition sensor subgroup; the scene perception sensor subgroup includes an ambient illuminance sensor, an integrated weather sensor, a roadside geomagnetic sensor, a video recognition camera and a road type identification sensor, which respectively collect ambient illuminance, weather conditions and visibility, vehicle passing frequency, vehicle quantity and speed, and road type identification information; the equipment state monitoring sensor subgroup includes a brightness feedback module, a drive circuit monitoring unit and a running time counter, which respectively collect real-time brightness of the street lamp, temperature and current-voltage fluctuation of the drive circuit, cumulative running time of the street lamp and related data of brightness attenuation; the operating parameter acquisition sensor subgroup includes a voltage sensor, a current sensor, a power factor acquisition module and an energy consumption monitoring unit, which respectively collect real-time voltage, real-time current, power factor and real-time energy consumption data during the operation of the street lamp.

3. The full spectrum LED street light system based on remote control and energy saving analysis according to claim 1, characterized in that, In the dual-mode collaborative communication module, the 5G / 6G communication link uses the URLLC slice technology, the slice bandwidth allocation range is 100Mbps to 1Gbps, the peak transmission rate is ≥10Gbps, and the end-to-end delay is ≤10ms; the scene and equipment state data upload uses a fragmentation transmission mechanism, the single package size is limited to 1KB to 10KB, retransmission is triggered when a fragment is lost, the retransmission number is ≤3 times, and the retransmission timeout time is 50ms.

4. The full spectrum LED street light system based on remote control and energy saving analysis of claim 1, wherein, In the dual-mode cooperative communication module, the PLC communication link adopts OFDM modulation and demodulation technology, the subcarrier spacing is 1kHz to 20kHz, the carrier frequency range is 2MHz to 30MHz, and the communication rate is 1Mbps to 10Mbps; the running parameter data transmission adopts frame structure + address coding mode, each frame contains target street lamp device address, data length, data content and check bit, and only the street lamp terminal with matched address receives and analyzes the data.

5. The full spectrum LED street light system based on remote control and energy saving analysis of claim 1, wherein, In the dual-mode cooperative communication module, the preset priority is divided according to the influence degree of the instruction on the system operation and the response timeliness requirement, and is divided into four levels from high to low: the first level is an emergency shutdown instruction; the second level is a spectrum adjustment instruction; the third level is a conventional dimming instruction; and the fourth level is a state query instruction; the first level instruction is forwarded through the 5G / 6G link, and the second to fourth level instructions are preferentially transmitted through the PLC link.

6. The full spectrum LED street light system based on remote control and energy saving analysis of claim 1, wherein, In the PINN constraint decision module, the physical information neural network model includes an input layer, a hidden layer, an output layer and a physical constraint layer: the input layer has 18 neurons corresponding to the scene, device and operation characteristics; the hidden layer has 3 to 5 layers, each layer has 64 to 256 neurons, and the activation function uses the ReLU function; the output layer has 6 neurons corresponding to the currents of four primary color LEDs, a dimming ratio and a switch state.

7. The remote control and energy saving analysis based full spectrum LED street light system according to claim 1, wherein, In the PINN constraint decision module, the specific process of calculating the predicted physical quantity using the built-in physical equation is as follows: first, extract the core parameters output by the model feature extraction layer, including... The luminous flux of each primary color LED Street light installation height, light emission angle of each primary color LED, and drive current.

1. Light source installation elevation angle; 2. Calculation of target points on the road surface using the law of superposition of illuminance. The initial illuminance value, corresponding to the mathematical expression is: ,in, Target locations on the road surface initial illuminance, These are the pre-set target monitoring points on the road surface. This refers to the number of primary color LEDs included in a full-spectrum LED street light. For the first Normal and position of each primary color LED light source The angle between directions, For the LED to the point Spatial distance, No. The luminous flux of a primary color LED The sphericity coefficient; the initial illuminance is corrected using Lambert's cosine law to obtain the point position. The corrected predicted illuminance is expressed as: ,in, Target location Predicted illuminance corrected by Lambert's cosine law Install elevation angle and position of light source The composite angle of azimuth; based on multiple road surface target points. Illumination uniformity can be predicted using the illumination uniformity calculation formula, which is: , Predicting lighting uniformity, It is the minimum corrected illuminance among multiple road surface target points. It is the arithmetic mean of the corrected illuminance at multiple road surface target points; combined with the human visual function, the tristimulus values ​​are obtained by weighting the spectra of each primary color LED. , , Substitute the values ​​into the color temperature calculation formula to predict the color temperature. The color temperature calculation formula is: , It predicts color temperature. It is a color temperature conversion function, and the predicted color rendering index is obtained based on the comparison results of the tristimulus values ​​and the CIE standard light source; simultaneously, it is based on the driving current of each primary color LED. and corresponding operating voltage , the power of the individual primary color LED is calculated , the formula is: , the predicted energy consumption value is obtained by summation, the formula is: , is the predicted total energy consumption of the full-spectrum LED street lamp, and finally a predicted physical quantity set containing the corrected predicted illuminance, predicted lighting uniformity, predicted color temperature, predicted color rendering index and predicted energy consumption value of each point is formed.

8. The full spectrum LED street light system based on remote control and energy saving analysis of claim 1, wherein, In the PINN constraint decision module, the multi-dimensional error includes data fitting error, physical law deviation, constraint satisfaction deviation and dynamic adaptation deviation, the data fitting error is the arithmetic mean of the absolute deviation of each control parameter in the preliminary control scheme and the corresponding parameters of the historical optimal sample set of the same scene, and the threshold is set to 0.04, the physical law deviation is the maximum value of the relative deviation of the predicted physical quantity and the theoretical calculation value of the built-in physical equation, and the threshold is set to 0.03, the constraint satisfaction deviation is the normalized deviation maximum value when the predicted physical quantity exceeds the preset constraint condition threshold, and the deviation value is 0 when the threshold is not exceeded, and the threshold is set to 0.02, and the dynamic adaptation deviation is the adaptation score deviation value calculated based on the scene real-time change rate and the historical same period change rule, the adaptation score range is 0-1, the deviation value is obtained by subtracting the adaptation score from 1, and the threshold is set to 0.

05.

9. The full spectrum LED street light system based on remote control and energy saving analysis of claim 1, wherein, The single lamp executes with the effect collection module, the LED drive circuit includes independent constant current drive unit and parameter collaborative control unit, each primary color LED corresponds the exclusive constant current drive unit, after the parameter collaborative control unit receives the drive current, voltage parameter in the standardization control instruction set, the accurate control signal is output to each constant current drive unit, the working current and voltage of each primary color LED are adjusted, the independent regulation and control of each primary color light flux are realized; The dimming signal is generated by the high-frequency signal generator built-in in the drive circuit through the PWM dimming technology, the dimming frequency is set to 30kHz, the duty ratio of the dimming signal is linearly adjusted in the range of 0% to 100% according to the dimming proportion in the control instruction, the duty ratio adjustment step is 0.1%, the duty ratio change rate is set through the gradual change control logic, the transition time between adjacent duty ratios is not less than 0.5s, the smooth change of the street lamp brightness is realized through the collaborative output of the parameter regulation and control of each primary color LED and the PWM dimming signal.

10. The full spectrum LED street light system based on remote control and energy saving analysis according to claim 1, characterized in that, The mathematical expression of the feedback weighted deviation optimization algorithm in the data feedback iterative optimization module is: ; wherein: is a comprehensive deviation value, is the total number of data items participating in the deviation analysis, is the actual value of the first feedback data in the data packet, is the first predicted data output by the PINN constraint decision module, is the reference benchmark value corresponding to the first data, is a minimum value, is the importance weight of the first data, and , is a device aging correction factor, is a scene dynamic coefficient, is a timeliness decay coefficient of the first data.

Images