Brightening illumination engineering brightness real-time monitoring regulation and control method and system based on big data

Abstract

The invention relates to a big-data-based real-time monitoring, regulation and control method and system for brightness of lighting engineering, and relates to the technical field of lighting engineering, and the method comprises the steps: transmitting the illumination intensity, people flow, traffic flow and weather data of a current lighting area to a cloud server; the cloud server inputs the illumination intensity, the pedestrian flow, the traffic flow and the weather data to an illumination demand prediction model; the illumination demand prediction model outputs an illumination demand prediction result; the cloud server obtains an illumination brightness regulation and control strategy based on the illumination demand prediction result and with energy saving as an optimization target; and the cloud server generates an illumination regulation and control instruction based on the illumination brightness regulation and control strategy and sends the illumination regulation and control instruction to a controller of the illumination equipment in the current illumination area so as to realize adjustment of the brightness of the illumination equipment.

Description

Technical Field

[0001] The present application relates to the technical field of lighting engineering, and in particular to a method and system for real-time monitoring and control of brightness of lighting engineering based on big data. Background Art

[0002] In traditional lighting projects, the brightness of lighting equipment is usually controlled by fixed settings or simple timing control. Fixed settings mean that the lighting brightness remains unchanged at any time, and although timing control can switch on or off or adjust the brightness according to the preset time, it is often only executed according to a fixed schedule and cannot flexibly respond to complex and changing actual environments.

[0003] Since it is impossible to adjust the lighting brightness in real time according to the actual light intensity, pedestrian flow, vehicle flow and weather conditions, excessive lighting often occurs. For example, in the middle of the night when there are very few pedestrians and vehicles, the lighting equipment still maintains a high brightness, resulting in a large amount of energy waste and increased operating costs. At the same time, during activities with large pedestrian flow or when the weather conditions are bad and natural light is insufficient, the lighting brightness may not provide sufficient lighting, affecting people's activity safety and visual experience; when there are few pedestrians and sufficient natural light, overly bright lighting will cause light pollution and interfere with the surrounding environment.

[0004] How to solve this problem is a technical problem that technicians in this field need to overcome. Summary of the invention

[0005] In order to at least partially solve the above technical problems, the present application provides a method and system for real-time monitoring and control of brightness of lighting projects based on big data.

[0006] In the first aspect, the present application provides a method for real-time monitoring and control of brightness of lighting projects based on big data, which adopts the following technical solution.

[0007] A method for real-time monitoring and control of brightness of lighting engineering based on big data, characterized by comprising:

[0008] Send the light intensity, pedestrian flow, vehicle flow and weather data of the current lighting area to the cloud server;

[0009] The cloud server inputs the light intensity, pedestrian flow, vehicle flow and weather data into the lighting demand prediction model; the lighting demand prediction model outputs the lighting demand prediction result;

[0010] The cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and with energy saving as the optimization goal;

[0011] The cloud server generates a lighting control instruction based on the lighting brightness control strategy and sends the lighting control instruction to the controller of the lighting equipment in the current lighting area to adjust the brightness of the lighting equipment.

[0012] By adopting the above technical solution, the present application dynamically adjusts the lighting brightness according to actual needs, and while meeting the lighting needs, it reduces energy consumption to the maximum extent, ensuring the lighting effect and saving energy.

[0013] Optionally, the cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and with energy saving as the optimization goal, including:

[0014] The cloud server receives the lighting demand prediction result output by the lighting demand prediction model, wherein the lighting demand prediction result includes lighting intensity demand information and lighting duration demand information;

[0015] The cloud server generates multiple candidate lighting brightness control schemes based on the lighting intensity demand information;

[0016] Calculate the expected energy consumption of each candidate lighting brightness control solution; the expected energy consumption is calculated based on the lighting duration, brightness, and energy consumption data at different brightness per unit time;

[0017] Compare the expected energy consumption of each candidate lighting brightness control scheme, and select the scheme with the lowest expected energy consumption and meeting the lighting demand prediction results as the final lighting brightness control strategy.

[0018] Optionally, the cloud server generates multiple candidate lighting brightness control schemes based on the lighting intensity demand information, including:

[0019] Determine a minimum lighting intensity requirement value and a maximum lighting intensity requirement value in the lighting intensity requirement information;

[0020] According to a predetermined brightness adjustment interval, a plurality of different brightness levels are divided between a minimum lighting intensity requirement value and a maximum lighting intensity requirement value;

[0021] Different lighting durations are configured for each divided brightness level; the lighting durations are determined based on the flow of people and vehicles;

[0022] Each brightness level and its corresponding lighting duration are combined to form multiple candidate lighting brightness control schemes.

[0023] Optionally, the method further includes:

[0024] Get the cumulative usage time of lighting equipment;

[0025] Obtaining an aging degree score of the device based on the accumulated usage time;

[0026] When the aging degree score is greater than the preset value, when generating the candidate lighting brightness control scheme based on the lighting intensity requirement information, for the lighting equipment with the aging degree score greater than the preset value, its maximum lighting intensity requirement value is reduced.

[0027] Optionally, for lighting devices with an aging degree score greater than a preset value, after reducing the maximum lighting intensity requirement value thereof, the method further includes:

[0028] Obtaining location information of lighting equipment with an aging score greater than a preset value;

[0029] Determine whether the aging equipment is concentrated in certain areas within the lighting project area;

[0030] For areas where aging equipment is concentrated, the maximum lighting intensity requirement values ​​of normal equipment and aging equipment in the entire area are reduced to reduce the difference in lighting intensity between the two.

[0031] For areas where aging equipment is dispersed, the brightness value that needs to be adjusted for surrounding normal equipment is calculated based on the maximum lighting intensity requirement value after the aging equipment is reduced, so that the surrounding normal equipment can adjust the brightness to achieve brightness transition;

[0032] During the lighting process, the brightness sensor is used to monitor the actual lighting intensity of the aging equipment and normal equipment in real time;

[0033] When the difference between the actual lighting intensity of the aged device and the actual lighting intensity of the surrounding normal devices is greater than a preset value, the lighting intensity of the aged device is increased, or the lighting intensity of the surrounding normal devices is reduced.

[0034] Optionally, the process of constructing the lighting demand prediction model includes:

[0035] Constructing a generator network; the generator network is used to predict the future lighting demand prediction results based on historical data; the generator network includes a convolution layer, an upsampling layer, a residual block and a skip connection module; the convolution layer is composed of multiple convolution layers stacked in sequence to extract features from historical data; the upsampling layer is used to increase the time step to predict the lighting demand at a future time point; the residual block is used to alleviate the gradient vanishing problem; the skip connection module is used to maintain the information transmission of low-level features during the upsampling process;

[0036] Constructing a discriminator network; the discriminator network is used to evaluate whether the prediction result of the generator network is consistent with the actual lighting demand status; the discriminator network includes: a convolution layer and a fully connected layer; the convolution layer is composed of multiple convolution layers arranged in sequence, and is used to extract features from the prediction result of the generator; the fully connected layer is used to integrate the features extracted by the convolution layer and to evaluate whether the prediction result is consistent with the actual lighting demand status;

[0037] Constructing a loss function module; the loss function module is used to quantify the difference between the generated lighting demand prediction and the actual lighting demand status; the loss function module includes a content loss module and an adversarial loss module; the content loss module is used to measure the difference between the generated lighting demand prediction and the actual lighting demand status; the adversarial loss module is used to encourage the generator to generate a prediction that is closer to the actual lighting demand status;

[0038] The initial model is obtained based on the constructed generator network, discriminator network and loss function module;

[0039] Use historical data to train the initial model and use the validation set to evaluate the prediction performance of the initial model and update the generator network and discriminator network during the validation process;

[0040] When the confidence of the initial model is greater than the preset value, the risk prediction model is obtained.

[0041] Optionally, the method further includes:

[0042] Assign a unique identification number including location information, model specifications to each lighting device;

[0043] Build communication links between lighting devices based on ZigBee in the lighting area;

[0044] When one of the lighting devices detects that its own current is abnormal, it immediately broadcasts the current data detected by itself and the location information generated based on its own ID to the adjacent devices through the local area communication network;

[0045] Lighting equipment with remote communication capabilities sends abnormal information and corresponding location information to the cloud.

[0046] Optionally, the method further includes:

[0047] Collect the current, voltage and actual light intensity of lighting equipment during operation and the difference between lighting control instructions;

[0048] Based on the machine learning algorithm, an equipment health status assessment model is built to evaluate the health status of lighting equipment; when the health status is less than the preset remaining usage time, a prompt message is sent to the operation and maintenance terminal.

[0049] In the second aspect, the present application provides a real-time monitoring and control system for brightness of lighting projects based on big data, which adopts the following technical solution.

[0050] A real-time monitoring and control system for brightness of lighting engineering based on big data, characterized by comprising:

[0051] The first processing module is used to: send the light intensity, pedestrian flow, vehicle flow and weather data of the current lighting area to the cloud server;

[0052] The second processing module is used for: the cloud server inputs the light intensity, pedestrian flow, vehicle flow and weather data into the lighting demand prediction model; the lighting demand prediction model outputs the lighting demand prediction result;

[0053] The third processing module is used for: the cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and taking energy saving as the optimization goal;

[0054] The fourth processing module is used for: the cloud server generates a lighting control instruction based on the lighting brightness control strategy and sends the lighting control instruction to the controller of the lighting equipment in the current lighting area to adjust the brightness of the lighting equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a method for real-time monitoring and controlling brightness of a lighting project based on big data according to an embodiment of the present application;

[0056] Figure 2 This is a system block diagram of a method for real-time monitoring and control of brightness of a lighting project based on big data in an embodiment of the present application;

[0057] In the figure, 201 is a first processing module; 202 is a second processing module; 203 is a third processing module; 204 is a fourth processing module. DETAILED DESCRIPTION

[0058] The following is combined with Figure 1-2 The present application is further described with specific embodiments:

[0059] The present application embodiment discloses a method for real-time monitoring and control of brightness of a lighting project based on big data, comprising the following steps:

[0060] Step 101: Send the light intensity, pedestrian flow, vehicle flow and weather data of the current lighting area to the cloud server.

[0061] Step 102: The cloud server inputs the light intensity, pedestrian flow, vehicle flow and weather data into a lighting demand prediction model; the lighting demand prediction model outputs a lighting demand prediction result.

[0062] Step 103: The cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and with energy saving as an optimization goal.

[0063] Step 104: The cloud server generates a lighting control instruction based on the lighting brightness control strategy and sends the lighting control instruction to the controller of the lighting device in the current lighting area to adjust the brightness of the lighting device.

[0064] Specifically, the light intensity, pedestrian flow, vehicle flow and weather data of the current lighting area are sent to the cloud server in real time. The cloud server processes these data using a lighting demand prediction model. The lighting demand prediction model is trained based on a large amount of historical data and is used to predict lighting demand. Based on this prediction result, the cloud server formulates a lighting brightness control strategy with energy saving as the optimization goal, avoiding the problem of over-lighting that often occurs in traditional lighting methods. The cloud server sends the lighting control instruction to the controller of the lighting equipment to adjust the brightness of the lighting equipment. Traditional lighting often uses fixed brightness or simple timing control, which cannot be adjusted according to real-time environmental changes and easily causes energy waste. This application dynamically adjusts the lighting brightness according to actual needs, while meeting lighting needs and minimizing energy consumption. The lighting effect is guaranteed and energy is saved.

[0065] As a specific implementation method of a real-time monitoring and control method for brightness of a lighting project based on big data, a cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and with energy saving as the optimization goal, including:

[0066] The cloud server receives the lighting demand prediction result output by the lighting demand prediction model, wherein the lighting demand prediction result includes lighting intensity demand information and lighting duration demand information;

[0067] The cloud server generates multiple candidate lighting brightness control schemes based on the lighting intensity demand information;

[0068] Calculate the expected energy consumption of each candidate lighting brightness control solution; the expected energy consumption is calculated based on the lighting duration, brightness, and energy consumption data at different brightness per unit time;

[0069] Compare the expected energy consumption of each candidate lighting brightness control scheme, and select the scheme with the lowest expected energy consumption and meeting the lighting demand prediction results as the final lighting brightness control strategy.

[0070] Specifically, the cloud server receives the prediction results containing the lighting intensity demand information and the lighting duration demand information, generates multiple candidate lighting brightness control schemes based on the lighting intensity demand information, and calculates the expected energy consumption of each candidate scheme according to the lighting duration, brightness, and energy consumption data at different brightness per unit time, thereby realizing energy consumption calculation; compares the expected energy consumption of each candidate scheme, and selects the scheme with the lowest expected energy consumption and that meets the lighting demand prediction results as the final strategy. Minimize energy consumption while meeting lighting needs. Compared with traditional lighting methods with fixed brightness or simple timing control, this method avoids energy waste, improves energy utilization efficiency, and saves a lot of operating costs for lighting projects.

[0071] As a specific implementation method of a real-time monitoring and control method for lighting engineering brightness based on big data, a cloud server generates multiple candidate lighting brightness control schemes based on lighting intensity demand information, including:

[0072] Determine a minimum lighting intensity requirement value and a maximum lighting intensity requirement value in the lighting intensity requirement information;

[0073] According to a predetermined brightness adjustment interval, a plurality of different brightness levels are divided between a minimum lighting intensity requirement value and a maximum lighting intensity requirement value;

[0074] Different lighting durations are configured for each divided brightness level; the lighting durations are determined based on the flow of people and vehicles;

[0075] Each brightness level and its corresponding lighting duration are combined to form multiple candidate lighting brightness control schemes.

[0076] Specifically, the minimum and maximum lighting intensity demand values ​​in the lighting intensity demand information are determined to ensure that the generated solution can meet the lighting demand and avoid insufficient or excessive brightness. According to the predetermined brightness adjustment interval, multiple different brightness levels are divided within this range, and different lighting durations are configured for each brightness level according to the flow of people and vehicles. Changes in the flow of people and vehicles directly affect the actual lighting needs. For example, when the flow of people and vehicles is large, higher brightness and longer lighting duration are required to ensure safety and comfort; when the flow of people and vehicles is small, the brightness can be reduced and the lighting duration can be shortened to save energy. Each brightness level and its corresponding lighting duration are combined to form multiple candidate solutions. By comparing these candidate solutions, the solution that best meets the actual needs and has the best energy-saving effect can be selected. While meeting the lighting needs, energy consumption is minimized and energy utilization efficiency is improved, achieving a balance between lighting effects and energy-saving goals.

[0077] As a specific implementation of a method for real-time monitoring and control of brightness of a lighting project based on big data, the method further includes:

[0078] Get the cumulative usage time of lighting equipment;

[0079] Obtaining an aging degree score of the device based on the accumulated usage time;

[0080] When the aging degree score is greater than the preset value, when generating the candidate lighting brightness control scheme based on the lighting intensity requirement information, for the lighting equipment with the aging degree score greater than the preset value, its maximum lighting intensity requirement value is reduced.

[0081] Specifically, the degree of aging of equipment is often closely related to its usage time. The longer the cumulative usage time, the higher the possibility and degree of equipment aging. The aging degree score of the equipment is obtained based on the cumulative usage time. When the aging degree score is greater than the preset value, the maximum lighting intensity requirement value of the aging equipment is reduced in the process of generating candidate lighting brightness control schemes. The performance of aging equipment has declined. If it is still operated at a high intensity, it may further accelerate its damage and even cause safety hazards. Lowering the maximum lighting intensity requirement value can reduce the load on the equipment and extend its service life. From an energy-saving perspective, the energy efficiency of aging equipment is usually low. Reducing its brightness requirement can reduce unnecessary energy consumption and achieve energy-saving goals. The above solution can take into account both equipment maintenance and energy saving while ensuring basic lighting needs.

[0082] As one implementation of a method for real-time monitoring and controlling brightness of a lighting project based on big data, for lighting equipment with an aging score greater than a preset value, after reducing its maximum lighting intensity requirement value, the method further includes:

[0083] Obtaining location information of lighting equipment with an aging score greater than a preset value;

[0084] Determine whether the aging equipment is concentrated in certain areas within the lighting project area;

[0085] For areas where aging equipment is concentrated, the maximum lighting intensity requirement values ​​of normal equipment and aging equipment in the entire area are reduced to reduce the difference in lighting intensity between the two.

[0086] For areas where aging equipment is dispersed, the brightness value that needs to be adjusted for surrounding normal equipment is calculated based on the maximum lighting intensity demand value after the aging equipment is reduced, so that the surrounding normal equipment can adjust the brightness to achieve brightness transition; during the lighting process, the brightness sensor is used to monitor the actual lighting intensity of the aging equipment and normal equipment in real time; when the difference between the actual lighting intensity of the aging equipment and the actual lighting intensity of the surrounding normal equipment is greater than the preset value, the lighting intensity of the aging equipment is increased, or the lighting intensity of the surrounding normal equipment is reduced.

[0087] Specifically, for areas where aging equipment is concentrated, the maximum lighting intensity requirement values ​​of normal equipment and aging equipment in the entire area are reduced at the same time to reduce the difference in lighting intensity between aging equipment and normal equipment, and avoid the phenomenon of uneven brightness in local areas. For areas where aging equipment is dispersed, the brightness of surrounding normal equipment is adjusted according to the reduced maximum lighting intensity requirement value of the aging equipment to achieve brightness transition and avoid sudden brightness changes, making the brightness changes in the lighting area more natural and smooth, thereby improving the overall quality of lighting. During the lighting process, the actual lighting intensity of the equipment is monitored in real time using a brightness sensor. When the difference in actual lighting intensity between the aging equipment and the surrounding normal equipment is greater than a preset value, adjustments are made by increasing the lighting intensity of the aging equipment or reducing the lighting intensity of the surrounding normal equipment, thereby ensuring that the brightness in the lighting area is always in a relatively balanced state and improving the lighting effect.

[0088] As one implementation method of a real-time monitoring and control method for brightness of a lighting project based on big data, the construction process of the lighting demand prediction model includes:

[0089] Constructing a generator network; the generator network is used to predict the future lighting demand prediction results based on historical data; the generator network includes a convolution layer, an upsampling layer, a residual block and a skip connection module; the convolution layer is composed of multiple convolution layers stacked in sequence to extract features from historical data; the upsampling layer is used to increase the time step to predict the lighting demand at a future time point; the residual block is used to alleviate the gradient vanishing problem; the skip connection module is used to maintain the information transmission of low-level features during the upsampling process;

[0090] Constructing a discriminator network; the discriminator network is used to evaluate whether the prediction result of the generator network is consistent with the actual lighting demand status; the discriminator network includes: a convolution layer and a fully connected layer; the convolution layer is composed of multiple convolution layers arranged in sequence, and is used to extract features from the prediction result of the generator; the fully connected layer is used to integrate the features extracted by the convolution layer and to evaluate whether the prediction result is consistent with the actual lighting demand status;

[0091] Constructing a loss function module; the loss function module is used to quantify the difference between the generated lighting demand prediction and the actual lighting demand status; the loss function module includes a content loss module and an adversarial loss module; the content loss module is used to measure the difference between the generated lighting demand prediction and the actual lighting demand status; the adversarial loss module is used to encourage the generator to generate a prediction that is closer to the actual lighting demand status;

[0092] The initial model is obtained based on the constructed generator network, discriminator network and loss function module;

[0093] Use historical data to train the initial model and use the validation set to evaluate the prediction performance of the initial model and update the generator network and discriminator network during the validation process;

[0094] When the confidence of the initial model is greater than the preset value, the risk prediction model is obtained.

[0095] Specifically, the generator network predicts the future lighting demand based on historical data (light intensity, pedestrian flow, vehicle flow and weather data). The convolution layer consists of multiple convolution layers stacked in sequence, and its core task is to extract features from historical data. In this lighting demand prediction model, historical data covers multi-dimensional information such as light intensity, pedestrian flow, vehicle flow and weather data at different time points. Changes in light intensity in different time periods: The convolution layer can automatically identify the periodic changes in light intensity over time within a day, such as the gradual increase in light intensity in the early morning and the gradual decrease in the evening; it can also capture the differences in light intensity in different seasons, such as the generally higher light intensity in summer than in winter. By learning these changes, the model can more accurately predict the impact of future light intensity on lighting demand. Peak and trough characteristics of pedestrian and vehicle flow: The convolution layer can detect the peak and trough of pedestrian and vehicle flow on weekdays and weekends, and at different times of the day. For example, during the morning and evening peak hours on weekdays, the pedestrian and vehicle flow is large, and the lighting demand may increase accordingly; while during the late night hours, the pedestrian and vehicle flow is small, and the lighting demand will also decrease. By learning these features, the model can more reasonably predict the lighting demand at different time points in the future. Features of weather data: The convolutional layer can extract features from weather data, such as the lighting characteristics under different weather conditions such as rain, cloudy, and sunny. When it rains or is cloudy, the natural light is insufficient and the lighting demand will increase; when it is sunny, the natural light is sufficient and the lighting demand will decrease.

[0096] The main function of the upsampling layer is to increase the time step in order to predict the lighting demand at future time points. In this model, the features extracted by the convolution layer are usually based on the data of the historical time step, and the upsampling layer expands these features in the time dimension through specific interpolation or convolution operations. Specifically, if the historical data is recorded hourly, the upsampling layer can expand these features to future hours or even finer time intervals (such as half an hour). By learning the time series patterns in the historical data, the upsampling layer can predict the lighting demand at different moments in the future based on the trend of past lighting demand changes. For example, if the light intensity has gradually weakened and the flow of people has gradually increased over the past few hours, the upsampling layer can predict that the lighting demand will increase in the future.

[0097] The residual block is used in this model to alleviate the gradient vanishing problem and enhance the feature learning ability. During the deep network training process, the gradient vanishing will make it difficult for the model to learn deep features.

[0098] In the lighting demand prediction model, as the number of network layers increases, information may be gradually lost during the transmission process, resulting in the model being unable to learn complex feature relationships. The residual block directly adds the input features to the features after the convolution operation through skip connections.

[0099] The role of the jump connection module in this model is to maintain the information transfer of low-level features during the upsampling process. During the upsampling process, low-level features may be lost due to changes in feature dimensions and information integration. In the lighting demand prediction model, low-level features may include short-term fluctuations in light intensity, instantaneous changes in pedestrian and vehicle traffic, etc. The jump connection module directly transfers the low-level features extracted by the previous convolutional layer to the subsequent upsampling layer. For example, a small fluctuation feature of light intensity at a certain moment is extracted in the early convolutional layer of the network. When the upsampling operation is performed, the feature is directly transferred to the corresponding position of the upsampling layer through the jump connection. This ensures that important low-level feature information is not omitted, so that the model can comprehensively consider these short-term and instantaneous change factors when predicting future lighting needs, thereby improving the accuracy of the prediction.

[0100] The discriminator network is used to evaluate whether the prediction results of the generator network are consistent with the actual lighting demand status. The convolutional layer consists of multiple convolutional layers arranged in sequence, which are used to extract features from the prediction results of the generator. In this model, the prediction results of the generator include the predicted values ​​of lighting demand at different time points in the future, and the convolutional layer extracts features from these predicted values. For example, it can identify the changing trend of the predicted lighting demand in the time series, such as whether there are abnormal peaks or troughs; it can also analyze the similarity between the predicted values ​​and the historical data patterns to determine whether the prediction results are reasonable. The fully connected layer integrates the features extracted by the convolutional layer and is used to evaluate whether the prediction results are consistent with the actual lighting demand status. The fully connected layer calculates a score based on the input features, which represents the probability that the discriminator believes that the lighting demand data generated by the generator is real data. If the score is close to 1, it means that the discriminator believes that the generated data is consistent with the actual lighting demand status; if the score is close to 0, it means that the discriminator believes that the generated data is false.

[0101] The loss function module is used to quantify the difference between the generated lighting demand prediction and the actual lighting demand situation. Content loss module: It is used to measure the difference between the generated lighting demand prediction and the actual lighting demand situation, and is calculated using the mean square error (MSE). The mean square error calculates the average of the squared differences between the generated lighting demand prediction value and the actual value. For example, for the lighting demand prediction value and the actual value at each time point, their difference is calculated and squared, and then the squared difference is averaged over all time points. This ensures that the model minimizes the deviation from the actual value when predicting the lighting demand. Adversarial loss module: Encourages the generator to generate predictions that are closer to the actual lighting demand situation, which is achieved through adversarial training with the discriminator. The goal of the generator is to make the discriminator judge its generated data as real data, while the goal of the discriminator is to accurately distinguish between generated data and real data. The adversarial loss adjusts the parameters of the generator based on the discriminator's judgment on the generated data, prompting the generator to generate more realistic lighting demand prediction results.

[0102] The generator network, the discriminator network, and the loss function module are combined to form a complete initial model. The model predicts future lighting demand through the generator, the discriminator evaluates the authenticity of the prediction results, and the loss function module guides the training and parameter adjustment of the model. Collect historical light intensity, pedestrian flow, vehicle flow, and weather data as well as the corresponding actual lighting demand data as training sets. These data should cover different seasons, different weather conditions, and different time periods to ensure that the model can learn comprehensive feature patterns. Input the training set data into the initial model for training, and gradually reduce the value of the loss function by continuously adjusting the parameters of the generator network and the discriminator network. During the training process, the generator will generate lighting demand prediction results based on historical data, the discriminator will evaluate these results, and then update the parameters of the generator and discriminator according to the feedback of the loss function. Use the validation set to evaluate the model during the training process. According to the evaluation results, such as the error between the predicted lighting demand and the actual value, the judgment accuracy of the discriminator, etc., update the parameters of the generator network and the discriminator network to improve the prediction performance of the model.

[0103] During the training and verification process, the confidence of the model is constantly evaluated. The confidence can be calculated in many ways, such as the error range between the predicted result and the actual value, the accuracy of the discriminator's judgment on the generated data, etc. When the confidence of the model is greater than the preset value, it is considered that the model has achieved good prediction performance, and the final lighting demand prediction model is obtained.

[0104] As one implementation of a method for real-time monitoring and controlling brightness of a lighting project based on big data, the method further includes:

[0105] Assign a unique identification number including location information, model specifications to each lighting device;

[0106] Build communication links between lighting devices based on ZigBee in the lighting area;

[0107] When one of the lighting devices detects that its own current is abnormal, it immediately broadcasts the current data detected by itself and the location information generated based on its own ID to the adjacent devices through the local area communication network;

[0108] Lighting equipment with remote communication capabilities sends abnormal information and corresponding location information to the cloud.

[0109] Specifically, a unique identification containing location information, model specifications is assigned to each lighting device; and a communication link between lighting devices is built based on ZigBee. When a lighting device detects an abnormal current in itself, it can immediately broadcast the current data and location information to adjacent devices through the local communication network. After the adjacent device receives the message, the lighting device with remote communication function sends the abnormal information and corresponding location information to the cloud, so that the cloud server can grasp the operating status of the equipment in the lighting system in real time, shortening the time for troubleshooting and saving hardware costs.

[0110] As one implementation of a method for real-time monitoring and controlling brightness of a lighting project based on big data, the method further includes:

[0111] Collect the current, voltage and actual light intensity of lighting equipment during operation and the difference between lighting control instructions;

[0112] Based on the machine learning algorithm, an equipment health status assessment model is built to evaluate the health status of lighting equipment; when the health status is less than the preset remaining usage time, a prompt message is sent to the operation and maintenance terminal.

[0113] The present application also provides a real-time monitoring and control system for brightness of lighting engineering based on big data, including:

[0114] The first processing module 201 is used to: send the light intensity, pedestrian flow, vehicle flow and weather data of the current lighting area to the cloud server;

[0115] The second processing module 202 is used for: the cloud server inputs the light intensity, pedestrian flow, vehicle flow and weather data into the lighting demand prediction model; the lighting demand prediction model outputs the lighting demand prediction result;

[0116] The third processing module 203 is used for: the cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and taking energy saving as the optimization goal;

[0117] The fourth processing module 204 is used for: the cloud server generates a lighting control instruction based on the lighting brightness control strategy and sends the lighting control instruction to the controller of the lighting device in the current lighting area to adjust the brightness of the lighting device.

[0118] It should be noted that the above embodiments are only used to illustrate the present application and are not intended to limit the technical solutions described in the present application. Although the present application has been described in detail in this specification with reference to the above embodiments, a person of ordinary skill in the art should understand that a person of ordinary skill in the art can still modify or make equivalent substitutions to the present application, and all technical solutions and improvements thereof that do not depart from the spirit and scope of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for real-time monitoring and control of brightness of lighting engineering based on big data, characterized in that: include: Send the light intensity, pedestrian flow, vehicle flow and weather data of the current lighting area to the cloud server; The cloud server inputs the light intensity, pedestrian flow, vehicle flow and weather data into the lighting demand prediction model; the lighting demand prediction model outputs the lighting demand prediction result; The cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and with energy saving as the optimization goal; The cloud server generates a lighting control instruction based on the lighting brightness control strategy and sends the lighting control instruction to the controller of the lighting equipment in the current lighting area to adjust the brightness of the lighting equipment.

2. According to the big data-based real-time monitoring and control method for lighting engineering brightness according to claim 1, it is characterized in that: The cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and with energy saving as the optimization goal, including: The cloud server receives the lighting demand prediction result output by the lighting demand prediction model, wherein the lighting demand prediction result includes lighting intensity demand information and lighting duration demand information; The cloud server generates multiple candidate lighting brightness control schemes based on the lighting intensity demand information; Calculate the expected energy consumption of each candidate lighting brightness control solution; the expected energy consumption is calculated based on the lighting duration, brightness, and energy consumption data at different brightness per unit time; Compare the expected energy consumption of each candidate lighting brightness control scheme, and select the scheme with the lowest expected energy consumption and meeting the lighting demand prediction results as the final lighting brightness control strategy.

3. According to the big data-based real-time monitoring and control method for lighting engineering brightness according to claim 2, it is characterized in that: The cloud server generates multiple candidate lighting brightness control schemes based on the lighting intensity demand information, including: Determine a minimum lighting intensity requirement value and a maximum lighting intensity requirement value in the lighting intensity requirement information; According to a predetermined brightness adjustment interval, a plurality of different brightness levels are divided between a minimum lighting intensity requirement value and a maximum lighting intensity requirement value; Different lighting durations are configured for each divided brightness level; the lighting durations are determined based on the flow of people and vehicles; Each brightness level and its corresponding lighting duration are combined to form multiple candidate lighting brightness control schemes.

4. According to the big data-based real-time monitoring and control method for lighting engineering brightness according to claim 3, it is characterized in that: The method further comprises: Get the cumulative usage time of lighting equipment; Obtaining an aging degree score of the device based on the accumulated usage time; When the aging degree score is greater than the preset value, when generating the candidate lighting brightness control scheme based on the lighting intensity requirement information, for the lighting equipment with the aging degree score greater than the preset value, its maximum lighting intensity requirement value is reduced.

5. According to the big data-based real-time monitoring and control method for lighting engineering brightness according to claim 4, it is characterized in that: For lighting devices whose aging degree score is greater than a preset value, after reducing the maximum lighting intensity requirement value thereof, the method further comprises: Obtaining location information of lighting equipment with an aging score greater than a preset value; Determine whether the aging equipment is concentrated in certain areas within the lighting project area; For areas where aging equipment is concentrated, the maximum lighting intensity requirement values ​​of normal equipment and aging equipment in the entire area are reduced to reduce the difference in lighting intensity between the two. For areas where aging equipment is dispersed, the brightness value that needs to be adjusted for surrounding normal equipment is calculated based on the maximum lighting intensity demand value after the aging equipment is reduced, so that the surrounding normal equipment can adjust the brightness to achieve brightness transition; during the lighting process, the brightness sensor is used to monitor the actual lighting intensity of the aging equipment and normal equipment in real time; when the difference between the actual lighting intensity of the aging equipment and the actual lighting intensity of the surrounding normal equipment is greater than the preset value, the lighting intensity of the aging equipment is increased, or the lighting intensity of the surrounding normal equipment is reduced.

6. The method for real-time monitoring and control of brightness of lighting engineering based on big data according to claim 5 is characterized in that: The construction process of the lighting demand prediction model includes: Constructing a generator network; the generator network is used to predict the future lighting demand prediction results based on historical data; the generator network includes a convolution layer, an upsampling layer, a residual block and a skip connection module; the convolution layer is composed of multiple convolution layers stacked in sequence to extract features from historical data; the upsampling layer is used to increase the time step to predict the lighting demand at a future time point; the residual block is used to alleviate the gradient vanishing problem; the skip connection module is used to maintain the information transmission of low-level features during the upsampling process; Constructing a discriminator network; the discriminator network is used to evaluate whether the prediction result of the generator network is consistent with the actual lighting demand status; the discriminator network includes: a convolution layer and a fully connected layer; the convolution layer is composed of multiple convolution layers arranged in sequence, and is used to extract features from the prediction result of the generator; the fully connected layer is used to integrate the features extracted by the convolution layer and to evaluate whether the prediction result is consistent with the actual lighting demand status; Constructing a loss function module; the loss function module is used to quantify the difference between the generated lighting demand prediction and the actual lighting demand status; the loss function module includes a content loss module and an adversarial loss module; the content loss module is used to measure the difference between the generated lighting demand prediction and the actual lighting demand status; the adversarial loss module is used to encourage the generator to generate a prediction that is closer to the actual lighting demand status; The initial model is obtained based on the constructed generator network, discriminator network and loss function module; Use historical data to train the initial model and use the validation set to evaluate the prediction performance of the initial model and update the generator network and discriminator network during the validation process; When the confidence of the initial model is greater than the preset value, the risk prediction model is obtained.

7. The method for real-time monitoring and control of brightness of lighting engineering based on big data according to claim 1 is characterized in that: The method further comprises: Assign a unique identification number including location information, model specifications to each lighting device; Build communication links between lighting devices based on ZigBee in the lighting area; When one of the lighting devices detects abnormal current in itself, it immediately broadcasts the current data detected by itself and the location information generated based on its own ID to the adjacent devices through the local area communication network; Lighting equipment with remote communication capabilities sends abnormal information and corresponding location information to the cloud.

8. The method for real-time monitoring and control of brightness of lighting engineering based on big data according to claim 7 is characterized in that: The method further comprises: Collect the current, voltage and actual light intensity of lighting equipment during operation and the difference between lighting control instructions; Based on the machine learning algorithm, an equipment health status assessment model is built to evaluate the health status of lighting equipment; when the health status is less than the preset remaining usage time, a prompt message is sent to the operation and maintenance terminal.

9. A real-time monitoring and control system for lighting engineering brightness based on big data, characterized in that: include: The first processing module is used to: send the light intensity, pedestrian flow, vehicle flow and weather data of the current lighting area to the cloud server; The second processing module is used for: the cloud server inputs the light intensity, pedestrian flow, vehicle flow and weather data into the lighting demand prediction model; the lighting demand prediction model outputs the lighting demand prediction result; The third processing module is used for: the cloud server obtains a lighting brightness control strategy based on the lighting demand prediction result and taking energy saving as the optimization goal; The fourth processing module is used for: the cloud server generates a lighting control instruction based on the lighting brightness control strategy and sends the lighting control instruction to the controller of the lighting equipment in the current lighting area to adjust the brightness of the lighting equipment.

Images