A method and system for optimizing tunnel lighting energy consumption based on ambient light perception

Abstract

This invention provides a method and system for optimizing tunnel lighting energy consumption based on ambient light perception, relating to the field of lighting control technology. It involves decomposing the temporal ambient light sequence into light wave inertia to obtain low-frequency light inertia components and high-frequency light impact components. A light field strain distribution map is constructed using the lighting unit grid, the low-frequency light inertia components, and the high-frequency light impact components. A compensation boundary is then determined based on the traffic flow prediction sequence and the light field strain distribution map. Within the compensation boundary, a multi-objective light energy redistribution is performed to obtain a light efficiency trade-off cluster. Based on this cluster, strain feedback values ​​for different lighting units are determined. Finally, a relaxation correction is applied to the multi-objective light energy redistribution process based on all strain feedback values. This invention enables decoupled perception of slow-changing trends and fast-changing disturbances in ambient light, and integrates traffic flow forecasting and relaxation correction of light field strain feedback to improve the energy consumption optimization capability of tunnel lighting under various operating conditions.

Description

Technical Field

[0001] This invention relates to the field of lighting control technology, and more specifically, to a method and system for optimizing tunnel lighting energy consumption based on ambient light perception. Background Technology

[0002] Lighting control technology is a crucial component of smart buildings and transportation infrastructure. Its core lies in dynamically adjusting the output luminous flux of luminaires based on ambient lighting conditions, human activity levels, and energy management needs to achieve a balance between visual comfort and energy efficiency. Tunnel lighting, as a special scenario within traffic lighting, experiences drastic changes in ambient light (the brightness at the tunnel entrance is determined by external natural light, while a smooth visual transition must be maintained inside the tunnel) and is strongly influenced by factors such as traffic flow, vehicle speed, and weather.

[0003] Ambient light perception-based tunnel lighting energy consumption optimization can dynamically adjust the lighting brightness of different tunnel sections by collecting ambient light brightness information at the tunnel entrance and inside the tunnel and combining it with traffic flow data, thereby reducing lighting energy consumption while ensuring traffic safety. However, existing tunnel lighting energy consumption optimization methods typically perform simple filtering or threshold judgment on measured ambient light values, failing to effectively decouple the slowly changing trend components and rapidly changing disturbance components in the ambient light signal. Furthermore, they rely on real-time traffic density feedback for lighting adjustment, lacking forward-looking prediction of future traffic flow changes. Spatially, they often employ segmented or regional dimming, without establishing a refined lighting unit grid and the light field coupling relationship between units, making it difficult to address model mismatch issues caused by long-term changes in the tunnel environment.

[0004] Therefore, how to decouple the perception of slow-changing trends and fast-changing disturbances in ambient light, and integrate traffic flow forecasting and relaxation correction of light field strain feedback to improve the energy consumption optimization capability of tunnel lighting under multiple working conditions is a challenge faced by the industry. Summary of the Invention

[0005] This invention provides a method and system for optimizing tunnel lighting energy consumption based on ambient light perception. It can achieve decoupled perception of slow-changing trends and fast-changing disturbances in ambient light, and integrate traffic flow forecasting and relaxation correction of light field strain feedback to improve the energy consumption optimization capability of tunnel lighting under multiple operating conditions.

[0006] In a first aspect, the present invention provides a method for optimizing tunnel lighting energy consumption based on ambient light perception, the optimization method comprising the following steps: Optical wave inertia decomposition was performed on the temporal ambient light sequence outside the tunnel to obtain low-frequency optical inertia components and high-frequency optical impact components. The interior of the tunnel is longitudinally discretized into a lighting unit grid, and a light field strain distribution map is constructed using the lighting unit grid, the low-frequency light inertia component, and the high-frequency light impact component. Traffic flow inside the tunnel is predicted to obtain a traffic flow prediction sequence. The compensation boundary is determined based on the traffic flow prediction sequence and the optical field strain distribution map. Multi-objective light energy redistribution is performed within the compensation boundary to obtain a light efficiency trade-off cluster. Gradient stepless light output is performed based on the light efficiency trade-off cluster to obtain strain feedback values ​​for different illumination units. The process of multi-target light energy redistribution is relaxed and corrected based on all strain feedback values ​​to obtain the energy consumption optimization weights for tunnel lighting.

[0007] In this embodiment, optical wave inertia decomposition is performed on the temporal ambient light sequence outside the tunnel to obtain low-frequency optical inertia components and high-frequency optical impact components, specifically including: A first-order inertial filter is applied to the temporal ambient light sequence outside the tunnel to obtain the low-frequency optical inertial component. The low-frequency optical inertia component in the temporal ambient light sequence outside the tunnel is separated and then truncated with a threshold to obtain the high-frequency optical impact component.

[0008] In this embodiment, discretizing the tunnel interior longitudinally into a lighting unit grid specifically includes: The lighting unit grid is obtained by dividing the tunnel longitudinally into lighting units with equal optical path intervals. Each lighting unit in the lighting unit grid has an optical inertia reference.

[0009] Furthermore, constructing the optical field strain distribution map using the illumination unit grid, the low-frequency optical inertia component, and the high-frequency optical impact component specifically includes: Calculate the luminance gradient tensor between adjacent lighting units in the lighting unit grid, and then propagate all luminance gradient tensors longitudinally to obtain a steady-state light field basis. The high-frequency light impact component is longitudinally diffused to obtain a transient optical field perturbation. The steady-state light field substrate and the transient light field perturbation are convolved to obtain the light strain coefficient of each lighting unit in the lighting unit grid. By arranging all the optical strain coefficients in a longitudinal order, the optical field strain distribution map is obtained.

[0010] Furthermore, traffic flow prediction is performed inside the tunnel, resulting in a traffic flow prediction sequence that specifically includes: Collect historical traffic flow time series data for the tunnel; The historical traffic flow time series is decomposed into multiple periods to extract hourly and daily periodic components. The hourly and daily periodic components are time-series predicted to obtain traffic flow prediction values, and then traffic flow prediction sequences are obtained.

[0011] Furthermore, determining the compensation boundary based on the traffic flow prediction sequence and the light field strain distribution map specifically includes: The compensation time window is obtained by filtering based on the traffic flow prediction sequence; Within the compensation time window, traverse the illumination units in the optical field strain distribution map where the optical strain coefficient exceeds the strain tolerance limit, and extract the vertical index interval; The compensation time window and the vertical index interval are fused by Cartesian product to obtain the compensation boundary.

[0012] Furthermore, performing multi-target light energy redistribution within the compensation boundary to obtain a light efficiency trade-off cluster specifically includes: Within the compensation boundary, a three-objective optimization function is constructed with the objectives of minimizing total light energy consumption, maximizing road surface illuminance uniformity, and minimizing visual flicker index. The three-objective optimization function is solved using a multi-objective evolutionary algorithm based on Pareto dominance to obtain a non-dominated solution set; Each solution in the non-dominated solution set is sorted from low to high light energy consumption to obtain a light efficiency trade-off cluster.

[0013] Furthermore, based on the aforementioned light efficiency trade-off cluster, gradient-based stepless light output is performed to obtain strain feedback values ​​for different illumination units, specifically including: Extract the target luminous flux of each illumination unit from the light efficiency trade-off cluster; According to the longitudinal order of each lighting unit, the luminous flux of all targets is differentially processed to obtain the luminous flux gradient constraint between adjacent lighting units; Based on the luminous flux gradient constraints, PWM dimming is executed unit by unit to drive the LED lamps to output continuously varying luminous flux. By monitoring continuously changing luminous flux and road surface reflectance, strain feedback values ​​of different lighting units can be obtained.

[0014] Furthermore, based on all strain feedback values, a relaxation correction is applied to the multi-target light energy redistribution process to obtain the energy consumption optimization weights for tunnel lighting, which specifically include: The strain feedback values ​​of all lighting units are arranged in a longitudinal order to form a strain feedback vector; The gradient descent method is used to iteratively adjust the light energy consumption weight, illuminance uniformity weight, and flicker index weight based on the strain feedback vector to obtain the energy consumption optimization weight for tunnel lighting.

[0015] Secondly, the present invention provides a tunnel lighting energy consumption optimization system based on ambient light perception, used to execute a tunnel lighting energy consumption optimization method based on ambient light perception, the optimization system comprising: The optical inertia analysis module is used to decompose the optical wave inertia of the temporal ambient light sequence outside the tunnel to obtain the low-frequency optical inertia component and the high-frequency optical impact component. The light field construction module is used to longitudinally discretize the interior of the tunnel into a lighting unit grid, and construct a light field strain distribution map through the lighting unit grid, the low-frequency light inertia component and the high-frequency light impact component. The boundary delineation module is used to predict traffic flow inside the tunnel, obtain a traffic flow prediction sequence, and determine the compensation boundary based on the traffic flow prediction sequence and the optical field strain distribution map. The gradient execution module is used to perform multi-target light energy redistribution within the compensation boundary to obtain a light efficiency trade-off cluster, and to perform gradient-type stepless light output based on the light efficiency trade-off cluster to obtain strain feedback values ​​of different illumination units. The relaxation correction module is used to perform relaxation correction on the multi-target light energy redistribution process based on all strain feedback values, so as to obtain the energy consumption optimization weight of tunnel lighting.

[0016] The technical solution provided by this invention has the following beneficial effects: This invention improves the energy consumption optimization capability of tunnel lighting under multiple operating conditions. Firstly, by performing optical wave inertia decomposition on the temporal ambient light sequence outside the tunnel, the ambient light signal is decoupled into a low-frequency optical inertia component representing diurnal periodic delay and a high-frequency optical impact component representing instantaneous optical disturbance, laying a physical foundation for differentiated control of slow-changing trends and fast-changing disturbances. Secondly, by longitudinally discretizing the tunnel interior into a lighting unit grid and constructing an optical field strain distribution map, the low-frequency optical inertia and high-frequency impact are mapped to each lighting unit, and the brightness gradient tensor between adjacent units is quantified, achieving structured modeling and strain sensing of the ambient light-spatial light field, providing a spatial constraint basis for refined dimming. Thirdly, by predicting traffic flow inside the tunnel to obtain a traffic flow prediction sequence, and combining this with the optical field strain distribution map, time windows and longitudinal index intervals are dynamically defined. The integrated compensation boundary enables proactive prediction and precise spatiotemporal positioning of lighting needs, overcoming the lag defects of traditional real-time feedback dimming. Then, by performing multi-objective light energy redistribution within the compensation boundary, a luminous efficacy trade-off cluster is obtained. Based on this cluster, gradient-based stepless light output is executed, achieving a Pareto optimal trade-off between energy consumption, illuminance uniformity, and visual flicker index. Simultaneously, luminous flux differential constraints ensure smooth transitions between adjacent units, enhancing multi-objective collaborative optimization capabilities. Finally, the multi-objective light energy redistribution process is relaxed and corrected based on the actual strain feedback values ​​of each lighting unit. The gradient descent method is used to iteratively adjust the optimization weights, enabling closed-loop learning and adaptive correction based on operational deviations. This enhances the system's robustness and continuous energy-saving effect under long-term changes such as luminaire light decay and sensor drift.

[0017] In summary, the technical solution adopted in this invention can achieve decoupled perception of the slow-changing trend and fast-changing disturbance of ambient light, and integrate traffic flow forecasting and relaxation correction of light field strain feedback to improve the energy consumption optimization capability of tunnel lighting under multiple working conditions. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a tunnel lighting energy consumption optimization method based on ambient light perception provided by the present invention; Figure 2 This is a schematic diagram of dividing the lighting unit grid according to the present invention; Figure 3 This is an exemplary flowchart for determining a traffic flow prediction sequence according to the present invention; Figure 4 This is a modular structure diagram of a tunnel lighting energy consumption optimization system based on ambient light perception, provided by the present invention. Detailed Implementation

[0020] The technical solutions of 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.

[0021] To better understand the technical solution of the present invention, the above technical solution will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] Example 1 refer to Figure 1 As shown in the figure, this is a flowchart of a tunnel lighting energy consumption optimization method based on ambient light perception according to this embodiment of the present invention. The optimization method includes the following steps: In step S1, the temporal ambient light sequence outside the tunnel is decomposed into optical wave inertia to obtain low-frequency optical inertia components and high-frequency optical impact components.

[0023] In this embodiment, optical wave inertia decomposition is performed on the temporal ambient light sequence outside the tunnel to obtain low-frequency optical inertia components and high-frequency optical impact components. Specifically, this can be achieved in the following manner: A first-order inertial filter is applied to the temporal ambient light sequence outside the tunnel to obtain the low-frequency optical inertial component. The low-frequency optical inertia component in the temporal ambient light sequence outside the tunnel is separated and then truncated with a threshold to obtain the high-frequency optical impact component.

[0024] It should be noted that, in this invention, the temporal ambient light sequence outside the tunnel refers to the time-series data of the natural light brightness outside the tunnel entrance, used to record the change process of the ambient light outside the tunnel over time. It can be collected by a photoelectric ambient light sensor installed on the outer wall of the tunnel entrance. The photoelectric ambient light sensor can convert the received natural light brightness into an electrical signal output. The acquisition frequency of the photoelectric ambient light sensor can be set between 25-100Hz, which is beneficial for capturing rapid light change events such as cloud cover, vehicle entry and exit, etc. The duration of the temporal ambient light sequence can be set to 10-30 minutes, which is beneficial for separating the physically meaningful low-frequency trend components and high-frequency disturbance components during light wave inertia decomposition.

[0025] In specific implementation, firstly, a first-order inertial filtering process is applied to the temporal ambient light sequence. The first-order inertial filtering process can perform a weighted average of the ambient light brightness value at the current moment and the filtered output value at the previous moment. The weight coefficient of the ambient light brightness value at the current moment is less than the weight coefficient of the filtered output value at the previous moment, which can make the filtered output value follow the change of the input value with a time lag. The sequence output after the first-order inertial filtering process is used as the low-frequency optical inertial component. The low-frequency optical inertial component can characterize the slowly changing low-frequency trend components in the ambient light. Finally, the ambient light brightness value at each moment in the time-series ambient light sequence is subtracted from the filtered output value of the low-frequency light inertia component at the corresponding moment to obtain a difference sequence. The absolute value of each difference in the difference sequence is taken and compared with a preset impact threshold. Differences exceeding the impact threshold are retained, and differences below the impact threshold are set to zero. The sequence obtained after threshold truncation is taken as the high-frequency light impact component. The high-frequency light impact component can characterize the transient disturbance component of drastic changes in ambient light. The impact threshold is used to distinguish between normal fluctuations and impact events that require a response. The value of the impact threshold can be set within the allowable range of brightness change rate in the tunnel lighting design specification.

[0026] It should be noted that the first-order inertial filtering in this invention is a low-pass filtering method that smooths the signal by weighting historical values ​​and current values. Its output exhibits an inertial delay characteristic in response to input changes. The low-frequency optical inertia component describes the slowly changing component with a large time constant in the ambient light signal, corresponding to long-term trends such as sunrise / sunset and gradual weather changes. The high-frequency optical impulse component describes instantaneous and violent fluctuations in the ambient light signal, such as rapidly passing clouds blocking sunlight or sudden changes in light caused by vehicles entering or exiting tunnels. By decomposing the temporal ambient light sequence into low-frequency optical inertia and high-frequency optical impulse components, differentiated processing of slowly changing trends and rapidly changing disturbances can be achieved.

[0027] In step S2, the interior of the tunnel is longitudinally discretized into a lighting unit grid, and a light field strain distribution map is constructed using the lighting unit grid, the low-frequency light inertia component, and the high-frequency light impact component.

[0028] In this embodiment, the longitudinal discretization of the tunnel interior into a lighting unit grid can be achieved in the following manner: The lighting unit grid is obtained by dividing the tunnel longitudinally into lighting units with equal optical path intervals. Each lighting unit in the lighting unit grid has an optical inertia reference.

[0029] In practice, firstly, multiple lighting units are divided longitudinally along the tunnel from the entrance to the exit at equal optical path intervals. Each lighting unit corresponds to a set of controllable tunnel lighting fixtures. All the divided lighting units are arranged in longitudinal order to form a lighting unit grid. Then, based on the cumulative optical path length of each lighting unit from the tunnel entrance, distance attenuation weighting is performed on the low-frequency optical inertia component. The farther the lighting unit is from the tunnel entrance, the smaller its weighting coefficient (i.e., the weight is assigned according to the proportion of the distance between the lighting unit and the tunnel entrance). The optical inertia value corresponding to each lighting unit after distance attenuation weighting is used as the optical inertia reference for that lighting unit.

[0030] It should be noted that, as Figure 2As shown in the figure, this is a schematic diagram of the division of lighting unit grids according to the present invention. In this embodiment, the lighting unit grid refers to an ordered set of basic lighting control units formed by discretizing the longitudinal space of the tunnel. Each lighting unit contains a set of physically adjacent luminaires, serving as the smallest control granularity for independent dimming. The equal optical path interval refers to dividing the tunnel longitudinally at equal intervals based on the length of the light propagation path. Since light propagation in the tunnel mainly occurs longitudinally, using an equal optical path interval ensures that adjacent lighting units have visually similar optical path differences, which is beneficial for a smooth brightness transition. The optical inertia reference refers to the optical inertia value corresponding to the basic light output that each lighting unit should maintain when unaffected by high-frequency disturbances. This reference is obtained by transferring the low-frequency optical inertia component through longitudinal attenuation, reflecting the steady-state mapping of the slowly changing trend of ambient light in different tunnel sections.

[0031] In this embodiment, the optical field strain distribution map constructed using the illumination unit grid, the low-frequency optical inertia component, and the high-frequency optical impact component can be implemented in the following manner: Calculate the luminance gradient tensor between adjacent lighting units in the lighting unit grid, and then propagate all luminance gradient tensors longitudinally to obtain a steady-state light field basis. The high-frequency light impact component is longitudinally diffused to obtain a transient optical field perturbation. The steady-state light field substrate and the transient light field perturbation are convolved to obtain the light strain coefficient of each lighting unit in the lighting unit grid. By arranging all the optical strain coefficients in a longitudinal order, the optical field strain distribution map is obtained.

[0032] In specific implementation, firstly, for every two adjacent lighting units in the lighting unit grid, the difference between the optical inertia reference of the preceding lighting unit and the optical inertia reference of the following lighting unit is calculated. This difference is divided by the longitudinal distance between the two lighting units to obtain the brightness gradient value between the adjacent lighting unit pairs. The brightness gradient values ​​of all adjacent lighting unit pairs are arranged in longitudinal order to form a one-dimensional sequence, which is used as the brightness gradient tensor. Secondly, each brightness gradient value in the brightness gradient tensor is passed sequentially from front to back along the tunnel longitudinal direction. That is, the steady-state light output value of each lighting unit is equal to the steady-state light output value of its preceding lighting unit plus the corresponding brightness gradient value. During the transmission process, the steady-state light output value of the first lighting unit is the value of the low-frequency optical inertia component. After the transmission is completed, the steady-state light output values ​​obtained by each lighting unit are arranged in longitudinal order to obtain the steady-state light field basis. Then, each impact value in the high-frequency light impact component is spatially diffused according to its longitudinal position within the bright cell grid. The diffusion rule is that each impact value affects the corresponding lighting cell and its adjacent lighting cells before and after it, with the degree of influence decreasing as the longitudinal distance increases. The light disturbance values ​​of all lighting cells affected by the impact are arranged in longitudinal order to obtain the transient light field disturbance. Finally, each steady-state light output value in the steady-state light field substrate is added element-wise to the disturbance value of the corresponding lighting cell in the transient light field disturbance to obtain the comprehensive light field value of each lighting cell. The difference between the comprehensive light field values ​​of adjacent lighting cells is calculated again and divided by the longitudinal distance. The calculation result is used as the light strain coefficient at the location of each lighting cell. The light strain coefficients of all lighting cells are arranged in longitudinal order, and the resulting one-dimensional sequence is used as the light field strain distribution map.

[0033] It should be noted that the luminance gradient tensor in this invention refers to a physical quantity describing the rate of change of the optical inertia reference between adjacent lighting units. It can represent the steepness of the luminance change along the longitudinal direction of the tunnel and is the basis for constructing the steady-state optical field substrate. The steady-state optical field substrate refers to the basic light output distribution of each lighting unit formed by the longitudinal propagation of only low-frequency optical inertia components. This distribution determines the basic luminance profile of tunnel lighting in the absence of high-frequency disturbances. Transient optical field disturbance refers to the instantaneous adjustment of the light output of each lighting unit formed by the diffusion of high-frequency light impact components along the longitudinal direction of the tunnel. This adjustment can simulate the spatial propagation and attenuation characteristics of external light abrupt changes inside the tunnel. The optical strain coefficient is the rate of change of the combined optical field value of adjacent lighting units. It is used to quantify the degree of local distortion of the luminance distribution at the location of each lighting unit. The larger the optical strain coefficient, the more drastic the luminance jump at that location. The optical field strain distribution map is a one-dimensional distribution map formed by arranging the optical strain coefficients of all lighting units in longitudinal order. It is used to characterize the optical field distortion state at various locations along the longitudinal direction of the tunnel.

[0034] In step S3, traffic flow inside the tunnel is predicted to obtain a traffic flow prediction sequence, and the compensation boundary is determined based on the traffic flow prediction sequence and the optical field strain distribution map.

[0035] Preferably, in this embodiment, reference Figure 3 As shown, this diagram is an exemplary flowchart for determining a traffic flow prediction sequence according to the present invention. In this embodiment, traffic flow prediction is performed inside the tunnel to obtain a traffic flow prediction sequence, which can be achieved through the following steps: First, in step S31, the historical traffic flow time series of the tunnel is collected; Then, in step S32, the historical traffic flow time series is decomposed into multiple periods to extract hourly periodic components and daily periodic components. Finally, in step S33, time-series prediction is performed on the hourly and daily periodic components to obtain traffic flow prediction values, and then a traffic flow prediction sequence is obtained.

[0036] In practice, firstly, historical traffic flow data is continuously collected by traffic flow detectors installed at the tunnel entrance and in various sections of the tunnel. The number of vehicles passing through at each sampling time is arranged chronologically to obtain a historical traffic flow time series. The data range for this historical traffic flow data can be from the past week to the present, which is beneficial for traffic flow prediction based on the traffic flow situation of the past week. Then, the historical traffic flow time series is decomposed into multiple periods. The sliding window averaging method is used to extract the repeating fluctuation component with a period of 1 hour to obtain the hourly periodic component. The sliding window averaging method with a period of 1 day is used to extract the repeating fluctuation component with a period of 24 hours to obtain the daily periodic component. During the decomposition, the random fluctuation part remaining after subtracting the hourly and daily periodic components from the historical traffic flow time series is treated as a residual term and is not used for subsequent prediction. Finally, the hourly and daily periodic components are input into the time series prediction network. The time series prediction network encodes and decodes the input sequence based on a long short-term memory structure and outputs the traffic flow prediction value for the future preset time step. After arranging all the traffic flow prediction values ​​in chronological order, a traffic flow prediction sequence is obtained. The unit of time step is hours. The preset time step can be preset according to the actual lighting environment of the tunnel, and the preset range can be set between 12 and 24 hours.

[0037] It should be noted that the historical traffic flow time series in this invention refers to the data set of traffic flow changes over time within the tunnel recorded by traffic flow detectors over a past period. Multi-period decomposition is the process of decomposing the original time series into periodic components and random residuals at different time scales. The hourly periodic component reflects hourly traffic flow patterns such as morning and evening peak hours, while the daily periodic component reflects daily traffic flow patterns such as weekdays and holidays. The time series prediction network is a neural network structure used to process time series data. This neural network structure includes memory units, can learn long-term dependencies in the sequence, and maps historical periodic components to future predicted values. This invention enables advance prediction of lighting needs through traffic flow forecasting, avoiding energy waste or insufficient brightness caused by response lag in real-time feedback methods.

[0038] In this embodiment, the compensation boundary is determined based on the traffic flow prediction sequence and the optical field strain distribution map, which can be achieved in the following manner: The compensation time window is obtained by filtering based on the traffic flow prediction sequence; Within the compensation time window, traverse the illumination units in the optical field strain distribution map where the optical strain coefficient exceeds the strain tolerance limit, and extract the vertical index interval; The compensation time window and the vertical index interval are fused by Cartesian product to obtain the compensation boundary.

[0039] In practice, firstly, the predicted traffic flow value at each predicted moment in the traffic flow prediction sequence is compared with a preset traffic flow density threshold. All consecutive moments where the predicted traffic flow value is greater than the traffic flow density threshold are selected. The start and end times of all these consecutive moments are considered as a time window, and all selected time windows are used as compensation time windows. The traffic flow density threshold can be set according to the tunnel's designed capacity. Then, within each compensation time window, each optical strain coefficient in the optical field strain distribution map is traversed. This optical strain coefficient is compared with the strain tolerance limit, and the vertical index numbers of all lighting units whose optical strain coefficients are greater than the strain tolerance limit are extracted. The minimum to maximum value among these vertical index numbers is used as the vertical index interval. The strain tolerance limit refers to the maximum allowable value of the optical strain coefficient in the optical field strain distribution map. Lighting units exceeding this limit pose a risk of excessive brightness jumps or visual discomfort and need to be included in the compensation range. Finally, each compensation time window and its corresponding vertical index interval are fused using a Cartesian product. That is, for each moment within each compensation time window and each lighting unit within the vertical index interval, a spatiotemporal point pair is generated. The set of all spatiotemporal point pairs is used as the compensation boundary.

[0040] It should be noted that the compensation time window in this invention refers to the time period during which dynamic light energy compensation needs to be activated. This time period consists of consecutive moments in the traffic flow prediction sequence that exceed the traffic flow density threshold. When the predicted traffic flow exceeds this threshold, it indicates that enhanced lighting is needed to ensure driving safety. The longitudinal index interval refers to the continuous numbering range of lighting units requiring compensation along the longitudinal direction of the tunnel, determined by the minimum and maximum indices of all exceeding-limit lighting units. Cartesian product fusion refers to combining the window in the time dimension with the interval in the spatial dimension to generate a spatiotemporal two-dimensional compensation boundary. Each spatiotemporal point within this boundary requires light energy redistribution. By fusing the compensation time window and the longitudinal index interval into a compensation boundary, this invention can accurately locate the spatiotemporal range requiring light energy compensation, thereby reducing computational burden and improving dimming efficiency.

[0041] In step S4, multi-target light energy redistribution is performed within the compensation boundary to obtain a light efficiency trade-off cluster. Gradient stepless light output is performed based on the light efficiency trade-off cluster to obtain strain feedback values ​​for different illumination units.

[0042] In this embodiment, performing multi-target light energy redistribution within the compensation boundary to obtain a light efficiency trade-off cluster can be specifically achieved in the following manner: Within the compensation boundary, a three-objective optimization function is constructed with the objectives of minimizing total light energy consumption, maximizing road surface illuminance uniformity, and minimizing visual flicker index. The three-objective optimization function is solved using a multi-objective evolutionary algorithm based on Pareto dominance to obtain a non-dominated solution set; Each solution in the non-dominated solution set is sorted from low to high light energy consumption to obtain a light efficiency trade-off cluster.

[0043] In practical implementation, firstly, at each spatiotemporal point within the compensation boundary, the output luminous flux of each lighting unit is used as the decision variable. A three-objective optimization function is constructed, with the three optimization objectives being minimizing the total light energy consumption of all lighting units, maximizing the road surface illuminance uniformity, and minimizing the visual flicker index. Here, the road surface illuminance uniformity is the ratio of the minimum illuminance to the average illuminance on the road surface, and the visual flicker index is the cumulative sum of the absolute values ​​of the luminous flux change rates at adjacent times. Then, a multi-objective evolutionary algorithm based on Pareto dominance is used to solve this three-objective optimization function. This algorithm generates an initial population containing multiple candidate solutions, each corresponding to a set of output luminous flux allocation schemes for each lighting unit. The population is iteratively updated by simulating selection, crossover, and mutation operations in biological evolution. After each generation, non-dominated solutions are retained according to the Pareto dominance relation; that is, no other solution is superior to this solution in all objectives and superior in at least one objective. After iterative convergence, the set of all non-dominated solutions is obtained as the non-dominated solution set. Finally, the total light energy consumption value corresponding to each solution in the non-dominated solution set is calculated, and all non-dominated solutions are sorted in order of total light energy consumption value from low to high. The sorted sequence is taken as the light efficiency trade-off cluster.

[0044] It should be noted that the total light energy consumption in this invention refers to the sum of electrical power consumption corresponding to the output luminous flux of all lighting units within the compensation boundary. The smaller this value, the better the energy-saving effect. Road surface illuminance uniformity is an important indicator for evaluating tunnel lighting quality. Higher uniformity indicates a more even distribution of road surface brightness and better visual comfort. The visual flicker index is used to quantify the drastic change in lighting brightness over time; a smaller flicker index indicates more stable light output. The Pareto dominance relation is a criterion used in multi-objective optimization to compare the merits of solutions. This criterion can be used to find a set of non-dominant compromise solutions, each with its own advantages for different objectives. The luminous efficacy trade-off cluster is a sequence formed by sorting non-dominant solutions by energy consumption. Each solution in this sequence represents a light energy allocation scheme and its corresponding three objective achievement values. In actual implementation, a suitable solution can be selected from this cluster based on the current degree of energy saving, uniformity, and flicker suppression to achieve gradient-based stepless light output.

[0045] In this embodiment, gradient-based stepless light output is performed based on the light efficiency trade-off cluster, and the strain feedback values ​​of different illumination units can be obtained in the following manner: Extract the target luminous flux of each illumination unit from the light efficiency trade-off cluster; According to the longitudinal order of each lighting unit, the luminous flux of all targets is differentially processed to obtain the luminous flux gradient constraint between adjacent lighting units; Based on the luminous flux gradient constraints, PWM dimming is executed unit by unit to drive the LED lamps to output continuously varying luminous flux. By monitoring continuously changing luminous flux and road surface reflectance, strain feedback values ​​of different lighting units can be obtained.

[0046] In practice, firstly, a solution is selected from the luminous efficacy tradeoff cluster. This solution corresponds to a set of target output luminous flux values ​​for each illumination unit. The target output luminous flux for each illumination unit is extracted from the vertical order of the illumination unit grid, thus obtaining the target luminous flux for each illumination unit. Then, according to the vertical order of the illumination units, the difference between the target luminous flux of two adjacent illumination units is calculated sequentially. This difference is used as the luminous flux gradient between these two illumination units. The sequence formed by arranging the luminous flux gradients of all adjacent units in vertical order is used as the luminous flux gradient constraint. Finally, based on the luminous flux gradient constraint, starting from the first lighting unit at the tunnel entrance, pulse width modulation dimming control is performed on each lighting unit sequentially towards the exit. That is, by adjusting the duty cycle of the pulse signal, the output luminous flux of the LED lamps in each lighting unit is changed, so that the difference in the actual output luminous flux of adjacent lighting units satisfies the luminous flux gradient value at the corresponding position in the luminous flux gradient constraint, thereby achieving a smooth light output that changes continuously from the entrance to the exit. After each dimming control cycle, the actual road surface reflection brightness of the corresponding road section of each lighting unit is collected by a brightness sensor installed on the road surface, and the actual output luminous flux of each lighting unit is recorded. The relative deviation between the actual output luminous flux of each lighting unit and the target luminous flux of that lighting unit is calculated, and this relative deviation is used as the strain feedback value of that lighting unit.

[0047] It should be noted that the target luminous flux in this invention refers to the ideal light output value that each lighting unit should achieve, as specified by the solution selected from the luminous efficacy trade-off cluster. This value is a compromise result obtained under multi-objective optimization. The luminous flux gradient refers to the rate of change of the target luminous flux between adjacent lighting units. Pulse width modulation (PWM) dimming is a technique that controls the average output light power of LED lamps by changing the ratio of the high-level duration to the period of the pulse signal. This technique enables continuous stepless adjustment of luminous flux. The strain feedback value is used to quantify the degree of deviation between the actual output and the target output of each lighting unit. This deviation may be caused by factors such as lamp light decay, sensor error, and nonlinearity of the drive circuit.

[0048] In step S5, the process of multi-target light energy redistribution is relaxed and corrected based on all strain feedback values ​​to obtain the energy consumption optimization weights for tunnel lighting.

[0049] In this embodiment, the process of multi-target light energy redistribution is relaxed and corrected based on all strain feedback values. The energy consumption optimization weights for tunnel lighting can be obtained in the following way: The strain feedback values ​​of all lighting units are arranged in a longitudinal order to form a strain feedback vector; The gradient descent method is used to iteratively adjust the light energy consumption weight, illuminance uniformity weight, and flicker index weight based on the strain feedback vector to obtain the energy consumption optimization weight for tunnel lighting.

[0050] In practice, firstly, the strain feedback values ​​of all lighting units are arranged sequentially according to the vertical order of the lighting unit grid to form a sequence as the strain feedback vector. Then, the values ​​of the light energy consumption weight, illuminance uniformity weight, and visual flicker index weight used in constructing the three-objective optimization function are obtained. The strain feedback vector is then subtracted element-wise from the predicted strain vector recorded in the previous round of light energy redistribution to obtain the residual vector. The sum of the squares of the residual vectors is used as the loss value. Finally, the gradient descent method is used to calculate the partial derivatives of the loss value with respect to the three parameters: light energy consumption weight, illuminance uniformity weight, and visual flicker index weight. This yields the gradient values ​​corresponding to these three weights. The value of each weight is then subtracted from the product of its gradient value and the preset learning step size to obtain the updated light energy consumption weight, updated illuminance uniformity weight, and updated visual flicker index weight. These three updated weights are used as the energy consumption optimization weights for tunnel lighting. The learning step size controls the magnitude of each parameter adjustment, and its value range is typically determined based on the specific application scenario and optimization objective. Preferably, considering that the light energy consumption weight, illuminance uniformity weight, and flicker index weight need to converge stably and avoid oscillations, the learning step size can be set between 0.001 and 0.1. A smaller learning step size allows for smoother weight updates, which is beneficial for fine-tuning but results in slower convergence; a larger learning step size can accelerate convergence but may cause weight overshoot. As an optional embodiment, the learning step size can be preset to 0.01. This value can balance convergence speed and stability in most scenarios. In practical applications, an adaptive learning step size strategy can also be adopted.

[0051] It should be noted that the strain feedback vector in this invention is a vector formed by arranging the actual output deviations of all lighting units in vertical order. This vector reflects the deviation distribution between the actual light output and the target light output at each position within the current control cycle. Relaxation correction refers to the gradual adjustment of the weight parameters of the optimization objective during multi-objective optimization using the deviation information from actual operation feedback, so that the optimization model gradually approaches the true optimal control boundary, similar to the relaxation iteration process in mechanics. Energy consumption optimization weight refers to the combination of optimization weights for the three objectives of light energy consumption, illuminance uniformity, and visual flicker index obtained after relaxation correction. This weight combination is used for the priority ranking of the luminous efficacy trade-off cluster in the next round of multi-objective light energy redistribution, which can realize closed-loop adaptive adjustment based on operational deviations, improving the energy-saving effect and control accuracy of the system in long-term operation.

[0052] In summary, the technical solution adopted in this invention can achieve decoupled perception of the slow-changing trend and fast-changing disturbance of ambient light, and integrate traffic flow forecasting and relaxation correction of light field strain feedback to improve the energy consumption optimization capability of tunnel lighting under multiple working conditions.

[0053] Example 2 This invention provides a tunnel lighting energy consumption optimization system based on ambient light sensing, with reference to... Figure 4 As shown, this figure is a modular structure diagram of a tunnel lighting energy consumption optimization system based on ambient light sensing according to the present invention. The optimization system includes: The optical inertia analysis module 100 is used to decompose the optical wave inertia of the temporal ambient light sequence outside the tunnel to obtain the low-frequency optical inertia component and the high-frequency optical impact component. The light field construction module 200 is used to longitudinally discretize the interior of the tunnel into an illumination unit grid, and construct a light field strain distribution map through the illumination unit grid, the low-frequency light inertia component and the high-frequency light impact component. The boundary delineation module 300 is used to predict traffic flow inside the tunnel, obtain a traffic flow prediction sequence, and determine the compensation boundary based on the traffic flow prediction sequence and the optical field strain distribution map. The gradient execution module 400 is used to perform multi-target light energy redistribution within the compensation boundary to obtain a light efficiency trade-off cluster, and perform gradient-type stepless light output based on the light efficiency trade-off cluster to obtain strain feedback values ​​of different illumination units. The relaxation correction module 500 is used to perform relaxation correction on the multi-target light energy redistribution process based on all strain feedback values ​​to obtain the energy consumption optimization weight of tunnel lighting.

[0054] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0055] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0056] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A method for optimizing energy consumption of tunnel lighting based on ambient light perception, characterized in that, Includes the following steps: Optical wave inertia decomposition was performed on the temporal ambient light sequence outside the tunnel to obtain low-frequency optical inertia components and high-frequency optical impact components. The interior of the tunnel is longitudinally discretized into a lighting unit grid, and a light field strain distribution map is constructed using the lighting unit grid, the low-frequency light inertia component, and the high-frequency light impact component. Traffic flow inside the tunnel is predicted to obtain a traffic flow prediction sequence. The compensation boundary is determined based on the traffic flow prediction sequence and the optical field strain distribution map. Multi-objective light energy redistribution is performed within the compensation boundary to obtain a light efficiency trade-off cluster. Gradient stepless light output is performed based on the light efficiency trade-off cluster to obtain strain feedback values ​​for different illumination units. The process of multi-target light energy redistribution is relaxed and corrected based on all strain feedback values ​​to obtain the energy consumption optimization weights for tunnel lighting.

2. The method for optimizing energy consumption of tunnel lighting based on ambient light perception according to claim 1, characterized in that, Optical wave inertia decomposition is performed on the temporal ambient light sequence outside the tunnel to obtain low-frequency optical inertia components and high-frequency optical impact components, specifically including: A first-order inertial filter is applied to the temporal ambient light sequence outside the tunnel to obtain the low-frequency optical inertial component. The low-frequency optical inertia component is separated and then truncated with a threshold to obtain the high-frequency optical impact component.

3. The method for tunnel lighting energy consumption optimization based on ambient light perception according to claim 1, characterized in that, The tunnel interior is longitudinally discretized into a lighting unit grid, specifically including: The lighting unit grid is obtained by dividing the tunnel longitudinally into lighting units with equal optical path intervals. Each lighting unit has an optical inertia reference.

4. The tunnel lighting energy consumption optimization method based on ambient light perception according to claim 1, characterized in that, A light field strain distribution map is constructed using the illumination unit grid, the low-frequency light inertia component, and the high-frequency light impact component, specifically including: Calculate the luminance gradient tensor between adjacent lighting units, and then propagate all luminance gradient tensors longitudinally to obtain the steady-state light field basis. The high-frequency light impact component is longitudinally diffused to obtain a transient optical field perturbation. The steady-state light field substrate and the transient light field perturbation are convolved to obtain the light strain coefficient of each lighting unit in the lighting unit grid. By arranging all the optical strain coefficients in a longitudinal order, the optical field strain distribution map is obtained.

5. The tunnel lighting energy consumption optimization method based on ambient light perception according to claim 1, characterized in that, Traffic flow prediction inside the tunnel is performed to obtain a traffic flow prediction sequence, which specifically includes: Collect historical traffic flow time series data for the tunnel; The historical traffic flow time series is decomposed into multiple periods to extract hourly and daily periodic components. The hourly and daily periodic components are time-series predicted to obtain traffic flow prediction values, and then traffic flow prediction sequences are obtained.

6. The tunnel lighting energy consumption optimization method based on ambient light perception according to claim 1, characterized in that, The compensation boundary is determined based on the traffic flow prediction sequence and the light field strain distribution map, specifically including: The compensation time window is obtained by filtering based on the traffic flow prediction sequence; Within the compensation time window, traverse the illumination units in the optical field strain distribution map where the optical strain coefficient exceeds the strain tolerance limit, and extract the vertical index interval; The compensation time window and the vertical index interval are fused by Cartesian product to obtain the compensation boundary.

7. The tunnel lighting energy consumption optimization method based on ambient light perception according to claim 1, characterized in that, Performing multi-objective light energy redistribution within the compensation boundary yields a light efficiency tradeoff cluster, specifically including: Within the compensation boundary, a three-objective optimization function is constructed with the objectives of minimizing total light energy consumption, maximizing road surface illuminance uniformity, and minimizing visual flicker index. The three-objective optimization function is solved using a multi-objective evolutionary algorithm based on Pareto dominance to obtain a non-dominated solution set; Each solution in the non-dominated solution set is sorted from low to high light energy consumption to obtain a light efficiency trade-off cluster.

8. The tunnel lighting energy consumption optimization method based on ambient light perception according to claim 1, characterized in that, Based on the aforementioned light efficiency tradeoff cluster, gradient-based stepless light output is performed to obtain strain feedback values ​​for different illumination units, specifically including: Extract the target luminous flux of each illumination unit from the light efficiency trade-off cluster; According to the longitudinal order of each lighting unit, the luminous flux of all targets is differentially processed to obtain the luminous flux gradient constraint between adjacent lighting units; Based on the luminous flux gradient constraints, PWM dimming is executed unit by unit to drive the LED lamps to output continuously varying luminous flux. By monitoring continuously changing luminous flux and road surface reflectance, strain feedback values ​​of different lighting units can be obtained.

9. The tunnel lighting energy consumption optimization method based on ambient light perception according to claim 1, characterized in that, The process of multi-target light energy redistribution is relaxed and corrected based on all strain feedback values ​​to obtain the energy consumption optimization weights for tunnel lighting, specifically including: The strain feedback values ​​of all lighting units are arranged in a longitudinal order to form a strain feedback vector; The gradient descent method is used to iteratively adjust the light energy consumption weight, illuminance uniformity weight, and flicker index weight based on the strain feedback vector to obtain the energy consumption optimization weight for tunnel lighting.

10. A tunnel lighting energy consumption optimization system based on ambient light perception, characterized in that, An optimization system for performing an ambient light sensing-based tunnel lighting energy consumption optimization method as described in any one of claims 1 to 9, the optimization system comprising: The optical inertia analysis module is used to decompose the optical wave inertia of the temporal ambient light sequence outside the tunnel to obtain the low-frequency optical inertia component and the high-frequency optical impact component. The light field construction module is used to longitudinally discretize the interior of the tunnel into a lighting unit grid, and construct a light field strain distribution map through the lighting unit grid, the low-frequency light inertia component and the high-frequency light impact component. The boundary delineation module is used to predict traffic flow inside the tunnel, obtain a traffic flow prediction sequence, and determine the compensation boundary based on the traffic flow prediction sequence and the optical field strain distribution map. The gradient execution module is used to perform multi-target light energy redistribution within the compensation boundary to obtain a light efficiency trade-off cluster, and to perform gradient-type stepless light output based on the light efficiency trade-off cluster to obtain strain feedback values ​​of different illumination units. The relaxation correction module is used to perform relaxation correction on the multi-target light energy redistribution process based on all strain feedback values, so as to obtain the energy consumption optimization weight of tunnel lighting.

Images