Solar street lamp brightness adjustment method based on digital twinning

Abstract

The application discloses a solar street lamp brightness adjustment method based on digital twinning, which comprises the following steps: S1, collecting multi-source data of the solar street lamp and preprocessing; S2, constructing a digital twin model, parameterizing modeling, and establishing a state update relationship; S3, forward expanding the model state to obtain a reachable state interval and an operation constraint boundary; S4, in the reachable state interval, introducing a control variable and discretizing, and recursively updating the state update relationship; S5, calculating the change of the state variable in the adjacent iteration step, and performing stability evaluation and constraint projection; S6, generating and issuing the brightness control result, and incrementally updating the state variable. The application can realize dynamic optimization control of the solar street lamp brightness based on the digital twin model driven by multi-source data, and improve the adaptability of brightness adjustment and energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of twin modeling technology, and in particular to a method for adjusting the brightness of solar streetlights based on digital twins. Background Technology

[0002] With the continuous advancement of smart city construction, solar streetlights, as a representative of green and low-carbon lighting, have been widely deployed in various urban public spaces. Traditional streetlight control strategies mostly rely on preset time periods, ambient light sensors, or set threshold triggering mechanisms, lacking the ability to dynamically respond to changes in streetlight operating status, energy utilization, and the environment. This can easily lead to problems such as insufficient illumination, wasted electricity, or accelerated equipment aging.

[0003] In recent years, some studies have begun to introduce state-aware and data-driven methods for intelligent street light control, such as using light sensors, battery power, and meteorological data to adjust brightness. However, most of these methods are based on static rules or shallow logic control, which makes it difficult to reflect the complex coupling relationships between variables within the system, effectively predict future operating states, and achieve stability constraints and dynamic corrections to control behavior.

[0004] Digital twin technology, with its ability to synchronize virtual and real systems, map states, and extrapolate evolution, is widely used in the modeling and control of complex systems such as industrial manufacturing and energy systems. However, its application in the control of solar streetlights is still immature, especially in areas such as state recursion, operational boundary constraints, control quantity search, and feedback closed-loop updates, where there are still significant technological gaps.

[0005] Therefore, how to provide a method for adjusting the brightness of solar streetlights based on digital twins is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose a method for adjusting the brightness of solar streetlights based on digital twins. This invention fully utilizes digital twin modeling, state recursive calculation, and constrained control search technology to model, deduce, and update the operating state of solar streetlights. It describes in detail the generation, stability assessment, and incremental correction process of the brightness control quantity under operating constraints, and has the advantages of predictable state evolution, constrained control process, high brightness adjustment stability, and strong operational adaptability.

[0007] A method for adjusting the brightness of a solar street light based on digital twins according to an embodiment of the present invention includes the following steps:

[0008] S1. Collect multi-source data from solar streetlights and preprocess it to form an operating status vector;

[0009] S2. Construct a digital twin model, map the running state vector to the model's state variables, parameterize the relationship between state variables in the digital twin model, and establish the state update relationship of state variables over time based on discrete time steps.

[0010] S3. The model state is expanded forward through a multi-step state recursion mechanism, and the range of values ​​of the state variables is constrained by the interval propagation method to obtain the reachable state interval and the running constraint boundary within the corresponding time interval.

[0011] S4. Within the reachable state range, a brightness control quantity is introduced as the control variable for the state update relationship. The control variable is discretized according to a preset step size, and the state update relationship is progressively recursively derived through a state search mechanism constrained by the operational boundary to form a state evolution trajectory.

[0012] S5. Iteratively expand the state evolution trajectory step by step, calculate the change in state variables in adjacent iteration steps, use the Lyapunov stability analysis method to evaluate stability, and perform constrained projection on the state variables.

[0013] S6. Based on the constrained state evolution trajectory, generate brightness control results, collect feedback data of the control results, and use an estimation update method to incrementally update the state variables.

[0014] Optionally, the preprocessing specifically includes:

[0015] The collected data on ambient light intensity, power generation, battery power, and brightness operation of solar streetlights are processed to perform timestamp alignment.

[0016] Anomaly detection methods were used to identify outliers in various types of time-aligned data, and outlier correction was completed through interpolation or neighborhood replacement.

[0017] Numerical scaling is performed on the data after anomaly correction;

[0018] The normalized data are concatenated and vectorized in chronological order to form a running status vector.

[0019] Optionally, the digital twin model includes a state representation layer, a state evolution layer, a control embedding layer, and a constraint expression layer, specifically including:

[0020] The state representation layer consists of multi-dimensional state variables, which are organized into a state space in the form of vectors at discrete time steps. Corresponding value range constraints and initial state descriptions are set for each state variable.

[0021] By extracting multidimensional indicators from the running state vector, a set of state variables is constructed, unique indices are assigned to establish a variable index mapping table, and the state variables are organized into a time series matrix based on discrete time steps. Initial values ​​and value ranges are set for each variable, forming a state space structure.

[0022] The state evolution layer describes the influence relationship between state variables through parameterized state update relationships, wherein the change process satisfies a predefined state transition form;

[0023] Based on the historical evolution trajectory of each variable in the state space, a bidirectional LSTM network is used to model the influence relationship between state variables, train and generate a set of state update functions, and use this set of functions as the dynamic evolution rule of state variables at time steps.

[0024] The control embedding layer is used to introduce brightness control quantities into the state update relationship. The brightness control quantities are control sequences that change with discrete time steps. The control sequences are composed of brightness level parameters, change step size parameters, and time sequence index parameters.

[0025] Construct a parameter set for the brightness control quantity, and insert the control sequence into the state update function through a parameter mapping function, using the brightness control quantity as an external input to participate in the state recursion process;

[0026] The constraint expression layer constructs a set of constraint conditions through logical constraint expressions and embeds the state update function in the form of a condition function to form a dynamic constraint mechanism for state evolution. The dynamic constraint mechanism includes constraints on the range of state variable values ​​and constraints on the boundary of state changes, and acts on the state update relationship in the form of constraint conditions.

[0027] Optionally, S2 specifically includes:

[0028] S21. Establish a set of state variables in the digital twin model, and assign variable indices corresponding to the dimensions of the running state vector to the set of state variables, and determine the index mapping relationship of the state variables.

[0029] S22. Based on the index mapping relationship, the running state vector is mapped to the state variables of the digital twin model, and the range constraint operator is used to apply boundary constraints to the state variables, specifically including:

[0030] According to the index mapping relationship, each dimension component in the running state vector is written into the corresponding state variable in sequence to form a state variable with the same dimension as the running state vector.

[0031] Configure the upper and lower boundaries of the value range for each state variable to obtain the set of boundary parameters corresponding to the state variable;

[0032] Define the range constraint operator as an interval projection operator, which performs the following constraint action on any state variable:

[0033] When the state variable is less than the lower boundary, the state variable is set to the lower boundary.

[0034] When the state variable is greater than the upper boundary, the state variable is set to the upper boundary.

[0035] The state variable remains unchanged when it is located between the upper and lower boundaries.

[0036] Perform consistency checks on the state variables after boundary constraints, and remove state variable terms that have illegal values ​​or undefined values.

[0037] S23. In the digital twin model, based on the mutual information measurement results between state variables, the correlation between state variables is determined, and the correlation weight between state variables is calculated.

[0038] S24. Based on the association weight, configure the update order and update weight of the state variables at discrete time steps, use a weighted combination method to aggregate the state variables of the previous time step, and introduce a state preservation term into the aggregation result to form the state update relationship of the state variables between adjacent time steps.

[0039] Optionally, S3 specifically includes:

[0040] S31. In the digital twin model, the values ​​of each state variable at the initial discrete time step are used as initial conditions to determine the number of discrete time steps within the preset time interval.

[0041] S32. In each discrete time step, the state variables are updated recursively using the explicit Euler method, specifically including:

[0042] At the current discrete time step, substitute the state variables from the previous discrete time step into the state update relation and calculate the state change corresponding to each state variable.

[0043] Based on the preset discrete time step, the state change is linearly superimposed with the state variable of the previous discrete time step to obtain the state variable of the current discrete time step.

[0044] The values ​​of each state variable at the current discrete time step are completely updated and used as the input state variables for the next discrete time step.

[0045] S33. At the initial discrete time step, for each state variable, construct the value perturbation range based on the change term of the state variable in the state update relationship, and form the corresponding initial interval state with the value of the state variable at the initial discrete time step as the center.

[0046] S34. In each subsequent discrete time step, the interval state of the previous discrete time step is progressively expanded using the interval arithmetic propagation method based on the state update relationship. The interval state of each state variable in the current discrete time step is calculated by performing addition and multiplication operations on the upper and lower boundaries of the interval respectively.

[0047] S35. In each discrete time step, the range constraint operator is executed on the obtained state variable interval states, and the constraint is transferred by restricting the upper and lower boundaries of the interval to the range of the state variables.

[0048] S36. Within a preset time interval, aggregate the interval states after constraint transfer at each discrete time step, take the minimum value of the lower boundary of the interval and the maximum value of the upper boundary of the interval for each state variable to form the reachable state interval within the corresponding time interval, and combine the upper and lower boundaries of the intervals of each discrete time step to form the running constraint boundary.

[0049] Optionally, S4 specifically includes:

[0050] S41. Within the reachable state range, the continuous value range of the brightness control quantity is discretized according to the preset step size to generate a discrete control set, and the brightness control quantity is combined with each state variable under the current discrete time step to form a state control node.

[0051] S42. Using the state control node as the search node, the search node is organized hierarchically according to the discrete time step, and the predecessor-successor mapping relationship of adjacent time layers is established through the state update relationship. At the same time, the state variable value of the next discrete time step corresponding to the brightness control quantity is calculated.

[0052] S43. In the hierarchical organization of search nodes, the search nodes are expanded layer by layer according to the Bellman forward recursion rule. During the recursive calculation, the running constraint boundary is checked for the state variable values ​​of each search node. When any state variable value does not meet the running constraint boundary, the subsequent expansion of the search node is terminated.

[0053] S44. For the search nodes that have passed the constraint boundary verification, perform path backtracking according to the predecessor-successor mapping relationship at each discrete time step, record the state variable values ​​and corresponding brightness control quantities at each discrete time step, and obtain the state evolution trajectory.

[0054] Optionally, the Bellman forward recursion rule specifically includes:

[0055] A hierarchical structure is constructed based on discrete time steps. The state control node under each time step is divided into corresponding hierarchical nodes, and the state control node of the previous time step is used as the set of predecessor nodes of the node in this layer.

[0056] In each time layer, based on the state update relationship, the state variables and brightness control quantities in the predecessor state control node are combined and input to calculate the corresponding candidate successor state control node.

[0057] During the generation of candidate successor nodes, the index information of the predecessor node is recorded, a mapping relationship between the predecessor node and the successor node is established, and the state control path is organized based on the mapping relationship.

[0058] Perform runtime constraint boundary checks on candidate successor state control nodes, eliminate nodes whose state variable values ​​exceed the boundary range, and use the remaining nodes as the recursive input set for the next time layer.

[0059] Optionally, S5 specifically includes:

[0060] S51. For the state variable group consisting of two consecutive discrete time steps in the state evolution trajectory, a sliding window structure is used for iterative expansion. In each sliding window, a state change vector is constructed based on the values ​​of the state variables in the consecutive time steps.

[0061] S52. Perform L2 norm operation on the state change vector to obtain the stability index sequence of the state variable change amplitude, and use the EWMA algorithm to smooth the index sequence.

[0062] S53. Perform monotonicity judgment on the smoothed stability index sequence. When the stability index shows a monotonically increasing trend in continuous time steps and the growth continues to exceed the set time step threshold, it is marked as an abnormal segment that does not meet the Lyapunov stability condition.

[0063] S54. For time steps determined not to meet the Lyapunov stability condition, perform perturbation projection adjustment on the corresponding state variables, specifically including:

[0064] Read the changes in each state variable at the current time step, and determine the disturbance offset of each state variable based on the changing trend of adjacent time steps;

[0065] The disturbance offsets are added one by one to the changes in the corresponding state variables to obtain the changes in the state variables after the disturbance.

[0066] Constraints are applied to the changes in state variables after disturbance, and the change components that exceed the upper and lower boundaries are truncated according to the corresponding boundary values.

[0067] Numerical compression mapping is performed on the changes in state variables after constraint clipping, and the changes are remapped to the continuous change range;

[0068] Based on the changes in the compressed state variables, the values ​​of the state variables at the current time step are updated and replaced.

[0069] Optionally, S6 specifically includes:

[0070] S61. In the digital twin model, the state evolution trajectory constrained by the operational constraint boundary is deordered, the state variables and corresponding brightness control quantities at each discrete time step are extracted, and the brightness control composite value under multiple paths is calculated using the interval weighted aggregation function to generate a brightness control result sequence.

[0071] S62. Input the brightness control result sequence into the execution interface module, encapsulate it into a control instruction set according to the unified instruction template, and send it to the corresponding terminal for execution;

[0072] S63. Collect the status data fed back by the terminal, match it according to the time step index in the control instruction set, compare it with the status variables of the corresponding time step in the digital twin model, and construct the feedback error vector.

[0073] S64. For the feedback error vector, the extended Kalman filter algorithm is used for state estimation update, specifically including:

[0074] Read the predicted and actual feedback values ​​of all state variables at the current time step, and calculate the state error increment;

[0075] A perturbation vector is constructed based on the direction of error increment, and a mapping rule is set between the upper limit of the perturbation amplitude and the direction of perturbation change to generate a perturbation correction factor.

[0076] Weighted modulation and nonlinear compression operations are performed on the perturbation correction factor to obtain the perturbation adjustment amount, which is then superimposed on the original predicted value to form candidate state variables;

[0077] Perform boundary pruning and consistency verification operations on the candidate state variables to complete the incremental update of the state variables.

[0078] The beneficial effects of this invention are:

[0079] First, by constructing a multi-layered digital twin model, the operating state of solar streetlights is represented in a structured manner as state variables. By combining parametric modeling methods with state update relationships, dynamic evolution modeling of the streetlight operating state in the time dimension is realized, which effectively improves the ability to characterize the state response under complex environmental changes.

[0080] Secondly, a state search mechanism based on Bellman forward recursion rules is adopted, which, combined with the discretized brightness control set and operational constraint boundary conditions, realizes the search for the optimal state evolution trajectory under constrained conditions, thereby enhancing the controllability and safety of the brightness adjustment process.

[0081] Finally, the incremental state update mechanism of Lyapunov stability analysis and extended Kalman filtering is introduced, which can quickly calibrate and dynamically correct the state feedback after the brightness control is executed, improve the consistency between the control results and the model prediction, and enhance the stability and robustness of the system in long-term operation. Attached Figure Description

[0082] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0083] Figure 1 This is an overall flowchart of a method for adjusting the brightness of a solar street light based on digital twins, as proposed in this invention.

[0084] Figure 2 This is a schematic diagram of the state recursion and interval propagation mechanism of a solar street light brightness adjustment method based on digital twin proposed in this invention;

[0085] Figure 3 This is a schematic diagram of the stability analysis and disturbance projection of a digital twin-based solar street light brightness adjustment method proposed in this invention. Detailed Implementation

[0086] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0087] refer to Figure 1-3 A method for adjusting the brightness of solar streetlights based on digital twins includes the following steps:

[0088] S1. Collect multi-source data from solar streetlights and preprocess it to form an operating status vector;

[0089] S2. Construct a digital twin model, map the running state vector to the model's state variables, parameterize the relationship between state variables in the digital twin model, and establish the state update relationship of state variables over time based on discrete time steps.

[0090] S3. The model state is expanded forward through a multi-step state recursion mechanism, and the range of values ​​of the state variables is constrained by the interval propagation method to obtain the reachable state interval and the running constraint boundary within the corresponding time interval.

[0091] S4. Within the reachable state range, a brightness control quantity is introduced as the control variable for the state update relationship. The control variable is discretized according to a preset step size, and the state update relationship is progressively recursively derived through a state search mechanism constrained by the operational boundary to form a state evolution trajectory.

[0092] S5. Iteratively expand the state evolution trajectory step by step, calculate the change in state variables in adjacent iteration steps, use the Lyapunov stability analysis method to evaluate stability, and perform constrained projection on the state variables.

[0093] S6. Based on the constrained state evolution trajectory, generate brightness control results, collect feedback data of the control results, and use an estimation update method to incrementally update the state variables.

[0094] In this embodiment, the preprocessing specifically includes:

[0095] The collected data on ambient light intensity, power generation, battery power, and brightness operation of solar streetlights are processed to perform timestamp alignment.

[0096] Anomaly detection methods were used to identify outliers in various types of time-aligned data, and outlier correction was completed through interpolation or neighborhood replacement.

[0097] Numerical scaling is performed on the data after anomaly correction;

[0098] The normalized data are concatenated and vectorized in chronological order to form a running status vector.

[0099] In this embodiment, the digital twin model includes a state representation layer, a state evolution layer, a control embedding layer, and a constraint expression layer, specifically including:

[0100] The state representation layer consists of multi-dimensional state variables, which are organized into a state space in the form of vectors at discrete time steps. Corresponding value range constraints and initial state descriptions are set for each state variable.

[0101] By extracting multidimensional indicators from the running state vector, a set of state variables is constructed, unique indices are assigned to establish a variable index mapping table, and the state variables are organized into a time series matrix based on discrete time steps. Initial values ​​and value ranges are set for each variable, forming a state space structure.

[0102] The state evolution layer describes the influence relationship between state variables through parameterized state update relationships, wherein the change process satisfies a predefined state transition form;

[0103] Based on the historical evolution trajectory of each variable in the state space, a bidirectional LSTM network is used to model the influence relationship between state variables, train and generate a set of state update functions, and use this set of functions as the dynamic evolution rule of state variables at time steps.

[0104] The control embedding layer is used to introduce brightness control quantities into the state update relationship. The brightness control quantities are control sequences that change with discrete time steps. The control sequences are composed of brightness level parameters, change step size parameters, and time sequence index parameters.

[0105] Construct a parameter set for the brightness control quantity, and insert the control sequence into the state update function through a parameter mapping function, using the brightness control quantity as an external input to participate in the state recursion process;

[0106] The constraint expression layer constructs a set of constraint conditions through logical constraint expressions and embeds the state update function in the form of a condition function to form a dynamic constraint mechanism for state evolution. The dynamic constraint mechanism includes constraints on the range of state variable values ​​and constraints on the boundary of state changes, and acts on the state update relationship in the form of constraint conditions.

[0107] In this embodiment, S2 specifically includes:

[0108] S21. Establish a set of state variables in the digital twin model, and assign variable indices corresponding to the dimensions of the running state vector to the set of state variables, and determine the index mapping relationship of the state variables.

[0109] S22. Based on the index mapping relationship, the running state vector is mapped to the state variables of the digital twin model, and the range constraint operator is used to apply boundary constraints to the state variables, specifically including:

[0110] According to the index mapping relationship, each dimension component in the running state vector is written into the corresponding state variable in sequence to form a state variable with the same dimension as the running state vector.

[0111] Configure the upper and lower boundaries of the value range for each state variable to obtain the set of boundary parameters corresponding to the state variable;

[0112] Define the range constraint operator as an interval projection operator, which performs the following constraint action on any state variable:

[0113] When the state variable is less than the lower boundary, the state variable is set to the lower boundary.

[0114] When the state variable is greater than the upper boundary, the state variable is set to the upper boundary.

[0115] The state variable remains unchanged when it is located between the upper and lower boundaries.

[0116] Perform consistency checks on the state variables after boundary constraints, and remove state variable terms that have illegal values ​​or undefined values.

[0117] S23. In the digital twin model, based on the mutual information measurement results between state variables, the correlation between state variables is determined, and the correlation weight between state variables is calculated.

[0118] S24. Based on the association weight, configure the update order and update weight of the state variables at discrete time steps, use a weighted combination method to aggregate the state variables of the previous time step, and introduce a state preservation term into the aggregation result to form the state update relationship of the state variables between adjacent time steps.

[0119] In this embodiment, S3 specifically includes:

[0120] S31. In the digital twin model, the values ​​of each state variable at the initial discrete time step are used as initial conditions to determine the number of discrete time steps within the preset time interval.

[0121] S32. In each discrete time step, the state variables are updated recursively using the explicit Euler method, specifically including:

[0122] At the current discrete time step, substitute the state variables from the previous discrete time step into the state update relation and calculate the state change corresponding to each state variable.

[0123] Based on the preset discrete time step, the state change is linearly superimposed with the state variable of the previous discrete time step to obtain the state variable of the current discrete time step.

[0124] The values ​​of each state variable at the current discrete time step are completely updated and used as the input state variables for the next discrete time step.

[0125] S33. At the initial discrete time step, for each state variable, construct the value perturbation range based on the change term of the state variable in the state update relationship, and form the corresponding initial interval state with the value of the state variable at the initial discrete time step as the center.

[0126] S34. In each subsequent discrete time step, the interval state of the previous discrete time step is progressively expanded using the interval arithmetic propagation method based on the state update relationship. The interval state of each state variable in the current discrete time step is calculated by performing addition and multiplication operations on the upper and lower boundaries of the interval respectively.

[0127] S35. In each discrete time step, the range constraint operator is applied to the obtained state variable intervals. The constraint is transferred by restricting the upper and lower boundaries of the intervals to the range of the state variables. Specifically, this includes:

[0128] For each state variable, read the corresponding lower boundary value and upper boundary value of the interval, and read the lower boundary and upper boundary of the value domain of the state variable as preset.

[0129] When the lower boundary value of the interval is less than the lower boundary value of the range, the lower boundary of the interval is updated to the lower boundary value of the range; otherwise, the lower boundary of the interval remains unchanged.

[0130] When the upper boundary value of the interval is greater than the upper boundary value of the range, the upper boundary of the interval is updated to the upper boundary value of the range; otherwise, the upper boundary of the interval remains unchanged.

[0131] Perform a consistency check on the interval state after updating the upper and lower boundaries. When the lower boundary of the interval is greater than the upper boundary of the interval, merge the lower boundary of the interval and the upper boundary of the interval into an equal value to obtain the corrected interval state variable.

[0132] S36. Within a preset time interval, aggregate the interval states after constraint transfer at each discrete time step, take the minimum value of the lower boundary of the interval and the maximum value of the upper boundary of the interval for each state variable to form the reachable state interval within the corresponding time interval, and combine the upper and lower boundaries of the intervals of each discrete time step to form the running constraint boundary.

[0133] In this embodiment, S4 specifically includes:

[0134] S41. Within the reachable state range, the continuous value range of the brightness control quantity is discretized according to the preset step size to generate a discrete control set, and the brightness control quantity is combined with each state variable under the current discrete time step to form a state control node.

[0135] S42. Using the state control node as the search node, the search node is organized hierarchically according to the discrete time step, and the predecessor-successor mapping relationship of adjacent time layers is established through the state update relationship. At the same time, the state variable value of the next discrete time step corresponding to the brightness control quantity is calculated.

[0136] S43. In the hierarchical organization of search nodes, the search nodes are expanded layer by layer according to the Bellman forward recursion rule. During the recursive calculation, the running constraint boundary is checked for the state variable values ​​of each search node. When any state variable value does not meet the running constraint boundary, the subsequent expansion of the search node is terminated.

[0137] S44. For the search nodes that have passed the constraint boundary verification, perform path backtracking according to the predecessor-successor mapping relationship at each discrete time step, record the state variable values ​​and corresponding brightness control quantities at each discrete time step, and obtain the state evolution trajectory.

[0138] In this embodiment, the Bellman forward recursion rule specifically includes:

[0139] A hierarchical structure is constructed based on discrete time steps. The state control node under each time step is divided into corresponding hierarchical nodes, and the state control node of the previous time step is used as the set of predecessor nodes of the node in this layer.

[0140] In each time layer, based on the state update relationship, the state variables and brightness control quantities in the predecessor state control node are combined and input to calculate the corresponding candidate successor state control node.

[0141] During the generation of candidate successor nodes, the index information of the predecessor node is recorded, a mapping relationship between the predecessor node and the successor node is established, and the state control path is organized based on the mapping relationship.

[0142] Perform runtime constraint boundary checks on candidate successor state control nodes, eliminate nodes whose state variable values ​​exceed the boundary range, and use the remaining nodes as the recursive input set for the next time layer.

[0143] In this embodiment, S5 specifically includes:

[0144] S51. For the state variable group consisting of two consecutive discrete time steps in the state evolution trajectory, a sliding window structure is used for iterative expansion. In each sliding window, a state change vector is constructed based on the values ​​of the state variables in the consecutive time steps.

[0145] S52. Perform L2 norm operation on the state change vector to obtain a stability index sequence of the state variable change amplitude, and smooth the index sequence using the EWMA algorithm, specifically including:

[0146] For the set of state variables formed by any two adjacent discrete time steps in the state evolution trajectory, a sliding window structure is set, and a state change vector is constructed in each window;

[0147] For each state change vector, a norm mapping is performed, and the L2 norm value is calculated using the Euclidean distance metric. The difference components of all state variables are squared, summed, and squared respectively to obtain the original stability index at the corresponding discrete time step.

[0148] Construct a time series of stability indices, map each L2 norm value to the corresponding time index in chronological order, generate an original index series with equal interval sampling, and mark local mutation points in the index series as abnormal fluctuations.

[0149] An exponentially weighted moving average algorithm is introduced into the stability index time series. An initial smoothing factor is set, and an exponential weight decreasing mechanism is implemented at each time step. The stability index at the current time step is weighted and summed with the previous smoothed value to generate a smoothed index series.

[0150] The smoothing index sequence is mapped to each discrete time step in the state evolution trajectory in a one-to-one correspondence manner, and the difference components before and after smoothing are retained in the sequence structure.

[0151] S53. Perform monotonicity judgment on the smoothed stability index sequence. When the stability index shows a monotonically increasing trend in continuous time steps and the growth continues to exceed the set time step threshold, it is marked as an abnormal segment that does not meet the Lyapunov stability condition.

[0152] S54. For time steps determined not to meet the Lyapunov stability condition, perform perturbation projection adjustment on the corresponding state variables, specifically including:

[0153] Read the changes in each state variable at the current time step, and determine the disturbance offset of each state variable based on the changing trend of adjacent time steps;

[0154] The disturbance offsets are added one by one to the changes in the corresponding state variables to obtain the changes in the state variables after the disturbance.

[0155] Constraints are applied to the changes in state variables after disturbance, and the change components that exceed the upper and lower boundaries are truncated according to the corresponding boundary values.

[0156] Numerical compression mapping is performed on the changes in state variables after constraint clipping, and the changes are remapped to the continuous change range;

[0157] Based on the changes in the compressed state variables, the values ​​of the state variables at the current time step are updated and replaced.

[0158] In this embodiment, S6 specifically includes:

[0159] S61. In the digital twin model, the state evolution trajectory constrained by the operational constraint boundary is deordered, the state variables and corresponding brightness control quantities at each discrete time step are extracted, and the brightness control composite value under multiple paths is calculated using the interval weighted aggregation function to generate a brightness control result sequence.

[0160] S62. Input the brightness control result sequence into the execution interface module, encapsulate it into a control instruction set according to the unified instruction template, and send it to the corresponding terminal for execution;

[0161] S63. Collect the status data fed back by the terminal, match it according to the time step index in the control instruction set, compare it with the status variables of the corresponding time step in the digital twin model, and construct the feedback error vector.

[0162] S64. For the feedback error vector, the extended Kalman filter algorithm is used for state estimation update, specifically including:

[0163] Read the predicted and actual feedback values ​​of all state variables at the current time step, and calculate the state error increment;

[0164] A perturbation vector is constructed based on the direction of error increment, and a mapping rule is set between the upper limit of the perturbation amplitude and the direction of perturbation change to generate a perturbation correction factor.

[0165] Weighted modulation and nonlinear compression operations are performed on the perturbation correction factor to obtain the perturbation adjustment amount, which is then superimposed on the original predicted value to form candidate state variables;

[0166] Perform boundary pruning and consistency verification operations on the candidate state variables to complete the incremental update of the state variables.

[0167] Example 1:

[0168] To verify the feasibility of this invention in practice, it was applied to a peri-urban area of ​​a city, where a large number of solar-powered streetlights are distributed. Due to the variable climate and unstable sunlight exposure in this area, traditional timed and brightness-based control strategies suffer from energy waste, insufficient brightness, and uneven lighting. Especially during the rainy season, some sections experience severe brightness decay or even malfunction. Traditional control systems rely on preset schedules or simplified models and cannot respond in real-time to changes in the streetlights' current energy state and external environment.

[0169] To address the aforementioned issues, this invention proposes a digital twin-based method for adjusting the brightness of solar streetlights. This method constructs a digital twin mapping between the streetlight's operating state and its control state, and dynamically adjusts the brightness control strategy by combining historical and real-time data, thereby achieving closed-loop optimization adjustment based on state evolution and control feedback.

[0170] During application, the system first collects multi-source data from each street light, including solar panel voltage, current, energy storage status, current brightness level, ambient illuminance, and pedestrian and vehicle traffic intensity. The data is then cleaned and normalized using wavelet denoising and anomaly removal algorithms to construct an operational state vector. This state vector is then mapped into a multi-layered digital twin model, which includes a state representation layer, a state evolution layer, a control embedding layer, and a constraint expression layer, supporting state recursion, control quantity injection, and operational constraint linkage.

[0171] In the model, a multi-step recursive approach to state variables is employed using the explicit Euler method, and an interval propagation algorithm is used to dynamically calculate the reachable state interval and operational boundary. Next, a discrete control set is constructed within the state space, and a Bellman forward recursive mechanism is used, combined with a predecessor-successor mapping relationship, to perform state search. Finally, the optimal state evolution trajectory is extracted within the constrained path, and the corresponding brightness control command is calculated based on the mapping relationship between control variables and states. In the feedback phase, an extended Kalman filter algorithm is used to incrementally correct the predicted state, further enhancing the model's robustness.

[0172] During three consecutive days of monitoring, the method was deployed on a 2.4-kilometer-long rural road, including 112 solar streetlights. Data shows that the method of this invention outperforms traditional strategies in maintaining brightness stability, saving energy consumption, and responding to environmental fluctuations. Particularly in terms of energy storage capacity management, dynamic brightness control effectively extends the lighting duration, preventing darkness at night. Furthermore, user feedback during actual use has been positive, indicating its strong practicality and promotional value in the fields of urban smart lighting and new energy utilization.

[0173] Table 1 Comparison of Streetlight Control Effects

[0174] project Traditional control strategies Method of the present invention Increase ratio Average brightness (Lux) 18.3 23.7 29.50% Average energy consumption (Wh / day) 140.6 116.2 -17.40% Energy storage capacity maintenance rate (%) 65.2 81.9 25.60% Number of abnormal brightness fluctuations (times / day) 9.2 3.1 -66.30% User satisfaction score (out of 10) 6.4 8.7 35.90%

[0175] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for adjusting the brightness of a solar street light based on digital twins, characterized in that, Includes the following steps: S1. Collect multi-source data from solar streetlights and preprocess it to form an operating status vector; S2. Construct a digital twin model, map the running state vector to the model's state variables, parameterize the relationship between state variables in the digital twin model, and establish the state update relationship of state variables over time based on discrete time steps. S3. The model state is expanded forward through a multi-step state recursion mechanism, and the range of values ​​of the state variables is constrained by the interval propagation method to obtain the reachable state interval and the running constraint boundary within the corresponding time interval. S4. Within the reachable state range, a brightness control quantity is introduced as the control variable for the state update relationship. The control variable is discretized according to a preset step size, and the state update relationship is progressively recursively derived through a state search mechanism constrained by the operational boundary to form a state evolution trajectory. S5. Iteratively expand the state evolution trajectory step by step, calculate the change in state variables in adjacent iteration steps, use the Lyapunov stability analysis method to evaluate stability, and perform constrained projection on the state variables. S6. Based on the constrained state evolution trajectory, generate brightness control results, collect feedback data of the control results, and use an estimation update method to incrementally update the state variables.

2. The method for adjusting the brightness of a solar street light based on digital twins according to claim 1, characterized in that, The preprocessing specifically includes: The collected data on ambient light intensity, power generation, battery power, and brightness operation of solar streetlights are processed to perform timestamp alignment. Anomaly detection methods were used to identify outliers in various types of time-aligned data, and outlier correction was completed through interpolation or neighborhood replacement. Numerical scaling is performed on the data after anomaly correction; The normalized data are concatenated and vectorized in chronological order to form a running status vector.

3. The method for adjusting the brightness of a solar street light based on digital twins according to claim 1, characterized in that, The digital twin model comprises a state representation layer, a state evolution layer, a control embedding layer, and a constraint expression layer, specifically including: The state representation layer consists of multi-dimensional state variables, which are organized into a state space in the form of vectors at discrete time steps. Corresponding value range constraints and initial state descriptions are set for each state variable. By extracting multidimensional indicators from the running state vector, a set of state variables is constructed, unique indices are assigned to establish a variable index mapping table, and the state variables are organized into a time series matrix based on discrete time steps. Initial values ​​and value ranges are set for each variable, forming a state space structure. The state evolution layer describes the influence relationship between state variables through parameterized state update relationships, wherein the change process satisfies a predefined state transition form; Based on the historical evolution trajectory of each variable in the state space, a bidirectional LSTM network is used to model the influence relationship between state variables, train and generate a set of state update functions, and use this set of functions as the dynamic evolution rule of state variables at time steps. The control embedding layer is used to introduce brightness control quantities into the state update relationship. The brightness control quantities are control sequences that change with discrete time steps. The control sequences are composed of brightness level parameters, change step size parameters, and time sequence index parameters. Construct a parameter set for the brightness control quantity, and insert the control sequence into the state update function through a parameter mapping function, using the brightness control quantity as an external input to participate in the state recursion process; The constraint expression layer constructs a set of constraint conditions through logical constraint expressions and embeds the state update function in the form of a condition function to form a dynamic constraint mechanism for state evolution. The dynamic constraint mechanism includes constraints on the range of state variable values ​​and constraints on the boundary of state changes, and acts on the state update relationship in the form of constraint conditions.

4. The method for adjusting the brightness of a solar street light based on digital twins according to claim 1, characterized in that, S2 specifically includes: S21. Establish a set of state variables in the digital twin model, and assign variable indices corresponding to the dimensions of the running state vector to the set of state variables, and determine the index mapping relationship of the state variables. S22. Based on the index mapping relationship, the running state vector is mapped to the state variables of the digital twin model, and the range constraint operator is used to apply boundary constraints to the state variables, specifically including: According to the index mapping relationship, each dimension component in the running state vector is written into the corresponding state variable in sequence to form a state variable with the same dimension as the running state vector. Configure the upper and lower boundaries of the value range for each state variable to obtain the set of boundary parameters corresponding to the state variable; Define the range constraint operator as an interval projection operator, which performs the following constraint action on any state variable: When the state variable is less than the lower boundary, the state variable is set to the lower boundary. When the state variable is greater than the upper boundary, the state variable is set to the upper boundary. The state variable remains unchanged when it is located between the upper and lower boundaries. Perform consistency checks on the state variables after boundary constraints, and remove state variable terms that have illegal values ​​or undefined values. S23. In the digital twin model, based on the mutual information measurement results between state variables, the correlation between state variables is determined, and the correlation weight between state variables is calculated. S24. Based on the association weight, configure the update order and update weight of the state variables at discrete time steps, use a weighted combination method to aggregate the state variables of the previous time step, and introduce a state preservation term into the aggregation result to form the state update relationship of the state variables between adjacent time steps.

5. The method for adjusting the brightness of a solar street light based on digital twins according to claim 1, characterized in that, S3 specifically includes: S31. In the digital twin model, the values ​​of each state variable at the initial discrete time step are used as initial conditions to determine the number of discrete time steps within the preset time interval. S32. In each discrete time step, the state variables are updated recursively using the explicit Euler method, specifically including: At the current discrete time step, substitute the state variables from the previous discrete time step into the state update relation and calculate the state change corresponding to each state variable. Based on the preset discrete time step, the state change is linearly superimposed with the state variable of the previous discrete time step to obtain the state variable of the current discrete time step. The values ​​of each state variable at the current discrete time step are completely updated and used as the input state variables for the next discrete time step. S33. At the initial discrete time step, for each state variable, construct the value perturbation range based on the change term of the state variable in the state update relationship, and form the corresponding initial interval state with the value of the state variable at the initial discrete time step as the center. S34. In each subsequent discrete time step, the interval state of the previous discrete time step is progressively expanded using the interval arithmetic propagation method based on the state update relationship. The interval state of each state variable in the current discrete time step is calculated by performing addition and multiplication operations on the upper and lower boundaries of the interval respectively. S35. In each discrete time step, the range constraint operator is executed on the obtained state variable interval states, and the constraint is transferred by restricting the upper and lower boundaries of the intervals to the range of the state variables. S36. Within a preset time interval, aggregate the interval states after constraint transfer at each discrete time step, take the minimum value of the lower boundary of the interval and the maximum value of the upper boundary of the interval for each state variable to form the reachable state interval within the corresponding time interval, and combine the upper and lower boundaries of the intervals of each discrete time step to form the running constraint boundary.

6. The method for adjusting the brightness of a solar street light based on digital twins according to claim 1, characterized in that, S4 specifically includes: S41. Within the reachable state range, the continuous value range of the brightness control quantity is discretized according to the preset step size to generate a discrete control set, and the brightness control quantity is combined with each state variable under the current discrete time step to form a state control node. S42. Using the state control node as the search node, the search node is organized hierarchically according to the discrete time step, and the predecessor-successor mapping relationship of adjacent time layers is established through the state update relationship. At the same time, the state variable value of the next discrete time step corresponding to the brightness control quantity is calculated. S43. In the hierarchical organization of search nodes, the search nodes are expanded layer by layer according to the Bellman forward recursion rule. During the recursive calculation, the running constraint boundary is checked for the state variable values ​​of each search node. When any state variable value does not meet the running constraint boundary, the subsequent expansion of the search node is terminated. S44. For the search nodes that have passed the constraint boundary verification, perform path backtracking according to the predecessor-successor mapping relationship at each discrete time step, record the state variable values ​​and corresponding brightness control quantities at each discrete time step, and obtain the state evolution trajectory.

7. The method for adjusting the brightness of a solar street light based on digital twins according to claim 6, characterized in that, The Bellman forward recursion rule specifically includes: A hierarchical structure is constructed based on discrete time steps. The state control node under each time step is divided into corresponding hierarchical nodes, and the state control node of the previous time step is used as the set of predecessor nodes of the node in this layer. In each time layer, based on the state update relationship, the state variables and brightness control quantities in the predecessor state control node are combined and input to calculate the corresponding candidate successor state control node. During the generation of candidate successor nodes, the index information of the predecessor node is recorded, a mapping relationship between the predecessor node and the successor node is established, and the state control path is organized based on the mapping relationship. Perform runtime constraint boundary checks on candidate successor state control nodes, eliminate nodes whose state variable values ​​exceed the boundary range, and use the remaining nodes as the recursive input set for the next time layer.

8. The method for adjusting the brightness of a solar street light based on digital twins according to claim 1, characterized in that, S5 specifically includes: S51. For the state variable group consisting of two consecutive discrete time steps in the state evolution trajectory, a sliding window structure is used for iterative expansion. In each sliding window, a state change vector is constructed based on the values ​​of the state variables in the consecutive time steps. S52. Perform L2 norm operation on the state change vector to obtain the stability index sequence of the state variable change amplitude, and use the EWMA algorithm to smooth the index sequence. S53. Perform monotonicity judgment on the smoothed stability index sequence. When the stability index shows a monotonically increasing trend in continuous time steps and the growth continues to exceed the set time step threshold, it is marked as an abnormal segment that does not meet the Lyapunov stability condition. S54. For time steps determined not to meet the Lyapunov stability condition, perform perturbation projection adjustment on the corresponding state variables, specifically including: Read the changes in each state variable at the current time step, and determine the disturbance offset of each state variable based on the changing trend of adjacent time steps; The disturbance offsets are added one by one to the changes in the corresponding state variables to obtain the changes in the state variables after the disturbance. Constraints are applied to the changes in state variables after disturbance, and the change components that exceed the upper and lower boundaries are truncated according to the corresponding boundary values. Numerical compression mapping is performed on the changes in state variables after constraint clipping, and the changes are remapped to the continuous change range; Based on the changes in the compressed state variables, the values ​​of the state variables at the current time step are updated and replaced.

9. The method for adjusting the brightness of a solar street light based on digital twins according to claim 1, characterized in that, S6 specifically includes: S61. In the digital twin model, the state evolution trajectory constrained by the operational constraint boundary is deordered, the state variables and corresponding brightness control quantities at each discrete time step are extracted, and the brightness control composite value under multiple paths is calculated using the interval weighted aggregation function to generate a brightness control result sequence. S62. Input the brightness control result sequence into the execution interface module, encapsulate it into a control instruction set according to the unified instruction template, and send it to the corresponding terminal for execution; S63. Collect the status data fed back by the terminal, match it according to the time step index in the control instruction set, compare it with the status variables of the corresponding time step in the digital twin model, and construct the feedback error vector. S64. For the feedback error vector, the extended Kalman filter algorithm is used for state estimation update, specifically including: Read the predicted and actual feedback values ​​of all state variables at the current time step, and calculate the state error increment; A disturbance vector is constructed based on the direction of the error increment, and a mapping rule is set between the upper limit of the disturbance amplitude and the direction of the disturbance change to generate a disturbance correction factor. Weighted modulation and nonlinear compression operations are performed on the perturbation correction factor to obtain the perturbation adjustment amount, which is then superimposed on the original predicted value to form candidate state variables; Perform boundary pruning and consistency verification operations on the candidate state variables to complete the incremental update of the state variables.

Images