A LED light source multi-light point projection module for environment rendering and control method thereof

Abstract

The present invention discloses an LED light source multi-light spot projection module for environment rendering and a control method thereof, which relates to the field of light source control technology, including collecting user needs and user browsing data through an app, and connecting the app with a light spot projector; constructing a control model, inputting user needs into the control model to obtain a control strategy, controlling the light spot projector according to the control strategy, and recording the control strategy as a user personalized plan; predicting user preferences according to the user personalized plan and user browsing data, and pushing them to the user through the app; the user remotely modifies the user needs to control the light spot projector through the app. The present invention constructs a control model to analyze user demand data to obtain a light spot projector control strategy, effectively meets user needs, provides accurate environment rendering effects, and predicts user personalized scenes according to user personalized plans and browsing data for push.

Description

Technical Field

[0001] The present invention relates to the technical field of light source control, and in particular to an LED light source multi-light point projection module for environment rendering and a control method thereof. Background Art

[0002] In recent years, with the rapid development of LED light source technology, its application in the fields of environmental rendering and atmosphere creation has been greatly promoted. LED light sources have gradually become the main light source choice in smart homes, commercial places and various creative scenes with their advantages of energy saving, high efficiency and strong controllability. Especially with the application of multi-light point projection technology, LED light sources can not only achieve precise brightness, color temperature and color control, but also create a variety of rich visual experiences through complex light arrangements and dynamic changes. Traditional LED light source control systems mostly use preset scene modes and manage light sources through remote controls or simple panel controls. It is difficult for users to deeply customize according to personal preferences, and it is impossible to fully utilize big data and artificial intelligence algorithms to accurately predict user behavior, resulting in limited user experience. Summary of the invention

[0003] In view of the problems existing in the above-mentioned existing LED light source multi-light point projection module for environment rendering and the control method thereof, the present invention is proposed.

[0004] Therefore, the problem to be solved by the present invention is that it is difficult for users to deeply customize according to their personal preferences, and it is impossible to fully utilize big data and artificial intelligence algorithms to accurately predict user behavior, resulting in limited user experience.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a control method of a LED light source multi-light spot projection module for environment rendering, which includes collecting user needs and user browsing data through an app, and connecting the app with a light spot projector;

[0006] Construct a control model, input user requirements into the control model to obtain a control strategy, control the light spot projector according to the control strategy, and record the control strategy as a user personalized solution;

[0007] Predict user preferences based on user personalization plans and user browsing data, and push them to users through the app;

[0008] Users can remotely modify user needs to control the light spot projector through the app.

[0009] As a preferred solution of the control method of the LED light source multi-light spot projection module for environment rendering described in the present invention, wherein: the collecting user needs and user browsing data through the app and connecting the app with the light spot projector refers to connecting the app with the light spot projector, collecting user demand data through the app, including light spot brightness, color, flashing frequency, quantity, distribution and projection distance, cleaning and filtering the user demand data after collecting the user demand data, and synchronously collecting the user browsing data in the app.

[0010] As a preferred solution of the control method of the LED light source multi-light spot projection module for environment rendering described in the present invention, wherein: the construction of the control model and inputting the user demand into the control model to obtain the control strategy include collecting the initial state data of the light spot projector through the app and preprocessing it, and using feature engineering to extract data features of the initial state data and user demand data;

[0011] The initial state data feature of the light spot projector is used as the initial state vector , and use the user demand data features as the ideal state vector of the light spot projector , based on the initial state vector and the ideal state vector Build the topology structure and convert the initial state vector and the ideal state vector The data types are used as layers, the data features in each data are used as nodes in the layers, the associations between nodes are used as edges, the node weights are set and the Pearson correlation coefficients between nodes are calculated as the node edge weights;

[0012] Define the local state tracking error in the topology :

[0013] ;in is the initial state vector The eigenvalue of the i-th node in the topological structure, is the ideal state vector The eigenvalue of the i-th node in the topological structure, n is the total number of nodes in the topological network;

[0014] Initialize the adaptive control gain parameters and construct the adaptive control law to generate the control input :

[0015] ;in , as well as Adjust , as well as The control gain parameters of

[0016] Solve the first positive definite matrix P through the Riccati equation and define the diagonal matrix as the second positive definite matrix , construct the Lyapunov function , define the optimization constraints as:

[0017] ;

[0018] ;

[0019] ;

[0020] ;in , as well as is the updated control gain parameter, T is the transposition operation;

[0021] Solve the Lyapunov function by optimizing the constraints to obtain the updated control gain parameters and bring them into the adaptive control law to obtain the updated control input ;

[0022] The consensus weight between nodes is calculated based on the node weight in the topology:

[0023] ; Where C is the consensus weight, m ​​is the number of topological structure levels, and n is the number of nodes in the topological structure. is the smoothing parameter, is the weight of the i-th node in the k-th level, is the weight of the jth node at the kth level;

[0024] Control input via consensus weights and updates The final control input Q is obtained:

[0025] ;in is the attenuation factor;

[0026] Select the RBF kernel function to construct the control model, set the initial kernel function parameters and regularization parameters, and define the control model input as the final control input Q, and the model output as the state vector of the light spot projector;

[0027] Define the model optimization objective function, use the model training set input to control the model for model iterative training, and label the model training set data;

[0028] In the iterative process, a parameter combination is formed for each kernel function parameter and regularization parameter within the defined parameter range, the Lagrange multiplier corresponding to each parameter combination is calculated, and the model optimization objective function value is calculated, and the parameter combination that minimizes the model optimization objective function value is selected as the optimal parameter combination, and its corresponding Lagrange multiplier is used as the model input;

[0029] The final control input Q is input into the control model to obtain the state vector of the light spot projector , define the objective function as:

[0030] ; Solve the objective function value and set the comparison threshold. When the objective function value is greater than the comparison threshold, the state vector of the light spot projector is As the initial state vector of the light spot projector, it is re-optimized until the objective function value is less than or equal to the comparison threshold, and then the optimization is stopped and the corresponding final control input Q is output;

[0031] The light spot projector control strategy is formed according to the corresponding final control input Q.

[0032] As a preferred solution of the control method of the LED light source multi-light spot projection module for environment rendering described in the present invention, wherein: the controlling the light spot projector according to the control strategy, and recording the control strategy as a user personalized plan means that the app obtains the control strategy of the light spot projector through the control model and controls the light spot projector to perform light spot projection according to the control strategy, the app records the control strategy and the light spot projection map, and forms a user personalized plan according to the user data and stores it in the user's personal database, allowing the user to upload the user personalized plan to the app cloud space for sharing the personalized plan, the app cloud space will display the shared personalized plan, and allow the user to browse and download the personalized plan from the app cloud space and switch to downloading the personalized plan with one click.

[0033] As a preferred solution of the control method of the LED light source multi-light point projection module for environment rendering of the present invention, wherein: the prediction of user preferences based on user personality schemes and user browsing data and the push to users through the app include:

[0034] Construct a convolutional neural network, which includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer;

[0035] Set the input layer input format to the user's personalized plan and user browsing data format, and connect it to the convolution layer;

[0036] The convolution layer is designed as a complex layer structure. Each convolution layer uses the ReLU function as the activation function, sets 32 filters, the filter size is 3*3, and the step size is 1;

[0037] The pooling layer is connected after the convolutional layer to perform maximum pooling on the convolutional layer output, setting the maximum pooling operation size to 2*2 and the step size to 2;

[0038] The fully connected layer uses the sigmoid function as the activation function to process the output of the pooling layer and outputs it through the output layer;

[0039] Use the neural network training set to train the convolutional neural network, annotate the data in the neural network training set, set the mean square error as the loss function, use the Adam optimizer to iteratively optimize the convolutional neural network parameters, calculate the error of the convolutional neural network in each iteration, and stop the iteration to obtain the optimal convolutional neural network parameters when the error no longer decreases significantly during continuous iterations;

[0040] The user's personalized plan and user browsing data are input into the convolutional neural network to obtain the user's personalized scene push, and the correlation between the user's personalized scene push and the user's personalized plan is calculated. A personalized scene push list is formed from high to low according to the correlation and pushed to the user through the app.

[0041] As a preferred solution of the control method of the LED light source multi-light spot projection module for environment rendering described in the present invention, wherein: the user remotely modifies the user needs to control the light spot projector through the app means that after the user modifies the user needs on the app, the app automatically forms the modified user demand data, and inputs it into the control model to obtain the control strategy to control the light spot projector, the app simultaneously collects the user voice data, converts the user voice data into text data based on voice recognition technology, and extracts the user needs in the text data through natural language processing technology to automatically form the modified user demand data.

[0042] Another object of the present invention is to provide an LED light source multi-light point projection module for environment rendering, which includes a light point projector and a processing module. The light point projector includes a light source, a reflector, a convex mirror and a scattering mirror. The light source emits light that is reflected by the reflector, passes through the convex mirror, and is then scattered into light spots by the scattering mirror.

[0043] As a preferred solution of the LED light source multi-light point projection module for environment rendering described in the present invention, wherein: the processing module includes:

[0044] Data collection module, used to collect and process user demand data and user browsing data;

[0045] The device control module is used to build a control model and obtain a control strategy to control the light spot projector according to the user demand data analysis;

[0046] Personalized push module, used to store user personalized plans and predict user personalized scenarios based on user personalized plans and user browsing data and push them to users through the app;

[0047] The remote control module is used for users to remotely modify user needs through the app to control the light spot projector and collect user voice to recognize user needs.

[0048] The beneficial effects of the present invention are as follows: the present invention collects user demand data, constructs a control model to analyze the user demand data, obtains a light spot projector control strategy, effectively meets user needs, provides accurate environmental rendering effects, and predicts user personalized scenes based on user personalized plans and browsing data for push, thereby improving user experience and personalization, and greatly improving user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0050] Figure 1 The present invention is a flow chart of a control method of an LED light source multi-light point projection module for environment rendering.

[0051] Figure 2 It is a structural schematic diagram of the light spot projector.

[0052] Figure 3 The module schematic diagram of the LED light source multi-light point projection module used for environment rendering. DETAILED DESCRIPTION

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0054] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0055] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0056] Embodiment 1:

[0057] Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and which provides a control method for a LED light source multi-light point projection module for environment rendering. The control method for a LED light source multi-light point projection module for environment rendering includes:

[0058] S1. Collect user needs and user browsing data through the app, and connect the app with the light spot projector;

[0059] Specifically, collecting user needs and user browsing data through the app and connecting the app to the light spot projector means connecting the app to the light spot projector, collecting user demand data through the app, including light spot brightness, color, flashing frequency, number, distribution and projection distance, and cleaning and filtering the user demand data after collecting it, and simultaneously collecting the user's browsing data in the app.

[0060] Gathering requirements through the APP not only improves the flexibility of the system, but also enables the light spot projection effect to be adjusted in real time according to user needs. The collection of user demand data is the basis for the subsequent generation of personalized control strategies, ensuring that the system can accurately meet the user's expected effects. Parameters such as the brightness, color and flashing frequency of the light spots can directly affect the construction of the environmental atmosphere and the user's visual experience, and the number and distribution of light spots determine the coverage range of the light spots and the complexity of the projection effect. By collecting and controlling these parameters, more complex and delicate light and shadow effects can be achieved. For example, controlling the distribution and color of multiple light spots can simulate complex light and shadow scenes such as the starry sky and the aurora; and the adjustment of the projection distance can adapt to physical spaces of different sizes and shapes to ensure the accurate projection of light spots. Through cleaning and filtering, the system can remove invalid or erroneous data and improve the accuracy of subsequent analysis and decision-making. For example, if a user accidentally clicks or Data that stays for a short time may interfere with the system's judgment. By filtering this data, the system can focus more on truly meaningful user behaviors. The cleaned data can ensure that the generation of personalized strategies is more accurate and reliable, reduce errors, and improve the matching degree of light spot projection effects. By analyzing the user's browsing behavior (such as dwell time, click preferences, etc.), the system can predict the light spot projection settings that the user may prefer in the future. This analysis provides data support for the dynamic adjustment of the light spot projector, enabling the system to optimize the light spot projection effect in advance without the user having to manually adjust it each time. For example, if the user frequently browses a specific light spot setting or scene, the system can push the scene without the user's active operation, thereby improving the smoothness and intelligence of the user experience. By connecting the APP to the light spot projector, the user can remotely control and adjust the status of the light spot in real time, realizing flexible operation anytime, anywhere.

[0061] S2. Build a control model, input user requirements into the control model to obtain a control strategy, control the light spot projector according to the control strategy, and record the control strategy as a user personalized solution;

[0062] Specifically, a control model is constructed and user requirements are input into the control model to obtain control strategies including:

[0063] The initial state data of the light spot projector is collected and preprocessed through the app, and feature engineering is used to extract data features of the initial state data and user demand data;

[0064] The initial state data feature of the light spot projector is used as the initial state vector , and use the user demand data features as the ideal state vector of the light spot projector , based on the initial state vector and the ideal state vector Build the topology structure and convert the initial state vector and the ideal state vector The data types are used as layers, the data features in each data are used as nodes in the layers, the associations between nodes are used as edges, the node weights are set and the Pearson correlation coefficients between nodes are calculated as the node edge weights;

[0065] Define the local state tracking error in the topology :

[0066] ;in is the initial state vector The eigenvalue of the i-th node in the topological structure, is the ideal state vector The eigenvalue of the i-th node in the topological structure, n is the total number of nodes in the topological network;

[0067] Initialize the adaptive control gain parameters and construct the adaptive control law to generate the control input :

[0068] ;in , as well as Adjust , as well as The control gain parameters of

[0069] Solve the first positive definite matrix P through the Riccati equation and define the diagonal matrix as the second positive definite matrix , construct the Lyapunov function :

[0070] , define the optimization constraints as:

[0071] ;

[0072] ;

[0073] ;

[0074] ;in , as well as is the updated control gain parameter, T is the transposition operation;

[0075] Solve the Lyapunov function by optimizing the constraints to obtain the updated control gain parameters and bring them into the adaptive control law to obtain the updated control input ;

[0076] The consensus weight between nodes is calculated based on the node weight in the topology:

[0077] ; Where C is the consensus weight, m ​​is the number of topological structure levels, and n is the number of nodes in the topological structure. is the smoothing parameter, is the weight of the i-th node in the k-th level, is the weight of the jth node at the kth level;

[0078] Control input via consensus weights and updates The final control input Q is obtained:

[0079] ;in is the attenuation factor;

[0080] Select the RBF kernel function to construct the control model, set the initial kernel function parameters and regularization parameters, and define the control model input as the final control input Q, and the model output as the state vector of the light spot projector;

[0081] Define the model optimization objective function, use the model training set input to control the model for model iterative training, and label the model training set data;

[0082] In the iterative process, a parameter combination is formed for each kernel function parameter and regularization parameter within the defined parameter range, the Lagrange multiplier corresponding to each parameter combination is calculated, and the model optimization objective function value is calculated, and the parameter combination that minimizes the model optimization objective function value is selected as the optimal parameter combination, and its corresponding Lagrange multiplier is used as the model input;

[0083] The final control input Q is input into the control model to obtain the state vector of the light spot projector , define the objective function as:

[0084] ; Solve the objective function value and set the comparison threshold. When the objective function value is greater than the comparison threshold, the state vector of the light spot projector is As the initial state vector of the light spot projector, it is re-optimized until the objective function value is less than or equal to the comparison threshold, and then the optimization is stopped and the corresponding final control input Q is output;

[0085] The light spot projector control strategy is formed according to the corresponding final control input Q.

[0086] Extracting the features of the initial state data and user demand data through feature engineering ensures the refinement and efficient processing of the data, so that the control model can better understand the user needs and the current state of the system, thereby improving the accuracy of the control strategy. The feature data extracted through feature engineering can significantly reduce the computational complexity of model training and prediction, and improve the real-time response capability of the system. By using the features of the initial state and the ideal state as nodes to build a topological structure, the system can intuitively represent and process the relationship between each state and demand in the form of a graph. This method can better capture the complex correlation between system states and ensure that all relevant factors are fully considered when generating control strategies. The local state tracking error is used to evaluate the difference between the current state and the ideal state, thereby guiding the adjustment of the adaptive control law. Through the feedback of the tracking error, the control strategy can be dynamically adjusted to ensure that the light spot projector always tends to meet the user's expected state. The adaptive control law combines the positive definite matrix solved by the Riccati equation to ensure the stability of the system in a dynamic environment. Through this method, the system can adjust the control strategy according to real-time feedback to avoid the occurrence of unstable states and improve control accuracy. Lyapunov The function is used to analyze and ensure the stability of the system. By optimizing the constraints to solve the updated control gain parameters, it ensures that the system is always in a stable state during operation and the tracking error is continuously reduced. This process not only improves the stability of the system, but also improves the effectiveness of the control strategy. The introduction of consensus weights ensures that the multi-level and multi-dimensional control systems can be coordinated when generating the final control strategy. By calculating the consensus weights, the system can balance the control inputs of each dimension and generate a more optimized final control input, thereby improving the overall control effect of the system. By selecting the RBF kernel function to construct the control model, the system can adapt to complex nonlinear inputs and ensure that the control strategy can achieve the best effect in various situations. Through model optimization, the system can find the optimal combination of parameters to make the control strategy more accurate and efficient. Through iterative training and optimization, the control model can continuously adjust and improve its performance. Each optimization iteration makes the model closer to the global optimum, so that the generated control strategy can better meet the control needs of the light spot projector.

[0087] Furthermore, controlling the light spot projector according to the control strategy and recording the control strategy as a user personalized plan means that the app obtains the light spot projector control strategy through the control model and then controls the light spot projector to perform light spot projection according to the control strategy. The app records the control strategy and the light spot projection map, and forms a user personalized plan based on the user data and stores it in the user's personal database, allowing the user to upload the user's personalized plan to the app cloud space for sharing the personalized plan. The app cloud space will display the shared personalized plan, and allow the user to browse and download the personalized plan from the app cloud space and switch to downloading the personalized plan with one click.

[0088] Precise control of the light spot projector is achieved through automated control strategies. The control strategies are generated based on the user's demand data to ensure that the light spot projection effect can meet the user's personalized needs. In specific applications, users can adjust the brightness, color and position of the light spot through the APP. The system will automatically execute these instructions and dynamically adjust the behavior of the light spot projector. This fine control method improves the user's operating convenience and experience, making the light spot effect more in line with the needs of different scenes. For example, in home theaters, smart lighting, stage performances and other scenes, users can flexibly adjust the lighting effect and record the control strategy as a user's personalized plan. Each user's light spot projection strategy can be independently stored according to their needs and called at any time when needed. This function not only improves the user experience of the system, but also increases the flexibility and scalability of personalization. Users can quickly switch their saved personalized plans in different scenes without resetting them. For example, users set the light spot layout that best suits their needs in certain specific environments. Then in the future In similar scenarios, you only need to call the plan with one click, which greatly saves time and operation steps. By storing the user's personalized plans in the personal database and allowing them to be uploaded to the APP's cloud space for sharing, the system expands the scope of use of personalized plans. The beneficial effect of this sharing mechanism is that it improves the openness and collaboration of the system. Users can not only enjoy the personalized light spot projection plans they set, but also browse and download plans shared by other users. The one-click switching function of the cloud space to display and download personalized plans enables users to more conveniently obtain and apply the personalized settings of other users. The beneficial effect of this function is that it greatly improves the convenience and usability of user operations. After the user downloads a personalized plan from the cloud space, he can directly apply it to his own light spot projector through the "one-click switching" function without complicated re-setting. This automated operation method makes the system more intelligent and efficient. Users only need simple operations to obtain ideal lighting effects, which is especially suitable for those users who are not very sensitive to technology.

[0089] S3, predict user preferences based on user personalization plans and user browsing data, and push them to users through the app;

[0090] Specifically, the user's preferences are predicted based on the user's personalized plan and user browsing data, and the app pushes information to the user, including:

[0091] Construct a convolutional neural network, which includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer;

[0092] Set the input layer input format to the user's personalized plan and user browsing data format, and connect it to the convolution layer;

[0093] The convolution layer is designed as a complex layer structure. Each convolution layer uses the ReLU function as the activation function, sets 32 filters, the filter size is 3*3, and the step size is 1;

[0094] The pooling layer is connected after the convolutional layer to perform maximum pooling on the convolutional layer output, setting the maximum pooling operation size to 2*2 and the step size to 2;

[0095] The fully connected layer uses the sigmoid function as the activation function to process the output of the pooling layer and outputs it through the output layer;

[0096] Use the neural network training set to train the convolutional neural network, annotate the data in the neural network training set, set the mean square error as the loss function, use the Adam optimizer to iteratively optimize the convolutional neural network parameters, calculate the error of the convolutional neural network in each iteration, and stop the iteration to obtain the optimal convolutional neural network parameters when the error no longer decreases significantly during continuous iterations;

[0097] The user's personalized plan and user browsing data are input into the convolutional neural network to obtain the user's personalized scene push, and the correlation between the user's personalized scene push and the user's personalized plan is calculated. A personalized scene push list is formed from high to low according to the correlation and pushed to the user through the app.

[0098] The convolutional neural network structure can process a large amount of complex user data, extract key features, and generate highly personalized scene push. The convolution layer is designed as a complex layer structure. Using 32 filters and a 3×3 convolution kernel size, it can efficiently capture important patterns in user personalized plans and browsing data. Each convolution layer uses the ReLU activation function to quickly and stably extract nonlinear features, thereby improving the expressiveness of the model. In addition, the maximum pooling layer further compresses the data dimension, retains core information while reducing the amount of calculation, thereby accelerating the network training and reasoning process. By formatting the preprocessed user personalized plans and browsing data and inputting them into the convolution layer, the neural network can more easily capture the correlation between the data. For example, the personalized plan can be It can contain information such as the color and brightness of light spots, while browsing data records the user's behavioral preferences. The convolution layer can automatically extract common patterns and preferences from these data through layer-by-layer learning, and provide support for subsequent scene push. This process simplifies the complexity of manual feature extraction, allowing the system to efficiently adapt to the needs of different users. After the convolution layer extracts features, the maximum pooling layer downsamples the data. The beneficial effect of this operation is that it reduces the interference of redundant information by reducing the dimension of the feature map, while enhancing the feature extraction capability. The 2×2 pooling size and step size of the pooling layer are set to 2 to ensure that the network's computational cost is reduced while retaining the main features. This combination enables the system to not only maintain high efficiency when processing large amounts of data, but also To ensure that the output personalized scene push has high accuracy, the fully connected layer uses the sigmoid function as the activation function. Its beneficial effect is that through nonlinear transformation, the features of the pooling layer output are mapped to the user preference space. The output of this layer is used as the final inferred user preference result and connected to the output layer to generate a personalized scene push list. The smoothness of the sigmoid function can ensure that the output result is between 0 and 1, making it easier for the network to predict the user's preference tendency, thereby providing more accurate scene recommendations. In the training process of the neural network, the model is iteratively optimized using the labeled training set data. By setting the mean square error (MSE) as the loss function and using the Adam optimizer for parameter optimization, it can be ensured. The network converges to the global optimal solution. The beneficial effect of this process is that by minimizing the prediction error, the convolutional neural network can more accurately infer the user's preferences, and then generate scene push notifications that are more in line with user needs. The dynamic learning rate adjustment mechanism of the Adam optimizer helps to increase the training speed, reduce the number of iterations, and ensure that the system obtains the optimal parameters within a reasonable time. Finally, the user's personalized plan and browsing data are input into the optimized convolutional neural network to generate personalized scene push notifications. The beneficial effect of this step is that the system can not only generate a push list based on the current user's data, but also calculate the relevance of each push notification with the user's personalized plan, thereby ensuring that the recommended scenes are highly personalized and targeted. For example, the system sorts according to relevance.Pushing the scenarios that are most consistent with the user's past preferences to the user allows the user to easily find the solution that best suits their needs among the many recommended scenarios, thus improving the personalized experience.

[0099] S4. The user remotely modifies the user's needs to control the light spot projector through the app;

[0100] Specifically, the user remotely modifies the user needs to control the light spot projector through the app, which means that after the user modifies the user needs on the app, the app automatically generates the modified user demand data, and inputs it into the control model to obtain the control strategy to control the light spot projector. The app simultaneously collects the user voice data, converts the user voice data into text data based on speech recognition technology, and extracts the user needs from the text data through natural language processing technology to automatically generate the modified user demand data.

[0101] The system can make dynamic adjustments based on the real-time needs of users to meet their personalized control needs. For example, users can adjust the brightness, color and projection mode of the light spot projector at any time through the APP without manually operating the device. This function greatly improves the flexibility and intelligence of the light spot projector, enabling it to adapt to a variety of usage scenarios, whether it is home lighting, commercial display, or performing arts lighting. When the user modifies the needs in the APP, the system will automatically generate new user demand data and input it into the control model. The beneficial effect of this step is that through the real-time generated data input, the system can quickly respond to changes in user needs to ensure that the projection effect of the light spot projector is consistent with user expectations. Through the control model, the system can generate the optimal light spot projection strategy based on the new data, thereby improving the accuracy of the projection effect. For example, after the user adjusts the brightness of the light spot, the control model will generate a new projection strategy based on environmental conditions and historical data to ensure that the brightness changes naturally. Voice interaction provides a more convenient and natural operation method, which improves the user experience. Through voice recognition technology, users can adjust the parameters of the light spot projector without manually entering complex commands. This convenient interactive method is suitable for various scenarios, especially those occasions where manual operation is not convenient. For example, when the user is doing other work, the lighting effect can be adjusted through simple voice commands. The combination of voice recognition technology and natural language processing technology can greatly improve the system's ability to understand user needs. In the present invention, voice recognition technology converts the user's voice data into text data, and NLP technology further analyzes the text, extracts the user's actual needs and generates corresponding modification data. The beneficial effect of this combination is that the system can not only recognize the user's voice commands, but also understand the intention behind the voice, thereby generating a more accurate control strategy. For example, when the user uses a voice command to "dim the lights", the system can automatically recognize and generate appropriate brightness adjustment commands. The control model can generate accurate control strategies based on different user demand data, so that the performance of the light spot projector is highly consistent with user expectations. The system can also be continuously optimized based on environmental feedback and historical data to ensure the continuous improvement of the light spot projection effect. For example, the system can continuously learn the user's preferences during the adjustment process and gradually generate a personalized solution that is more in line with user habits.

[0102] Furthermore, the light spot projector includes a light source, a reflector, a convex mirror and a scattering mirror. The light source emits light which is reflected by the reflector, passes through the convex mirror and is finally scattered into a light spot by the scattering mirror. The light density of the light source is greater than 700 lm per square millimeter, and the LED chip uses a light source with a light-emitting area of ​​less than 1 square millimeter.

[0103] Embodiment 2:

[0104] Reference Figure 3, which is the second embodiment of the present invention. This embodiment is different from the previous embodiment and provides an LED light source multi-light point projection module for environment rendering, which includes a light point projector and a processing module. The light point projector includes a light source, a reflector, a convex mirror and a scattering mirror. The light source emits light that is reflected by the reflector, passes through the convex mirror, and is then scattered into light spots by the scattering mirror.

[0105] Specifically, the processing module includes:

[0106] Data collection module, used to collect and process user demand data and user browsing data;

[0107] The device control module is used to build a control model and obtain a control strategy to control the light spot projector according to the user demand data analysis;

[0108] Personalized push module, used to store user personalized plans and predict user personalized scenarios based on user personalized plans and user browsing data and push them to users through the app;

[0109] The remote control module is used for users to remotely modify user needs through the app to control the light spot projector and collect user voice to recognize user needs.

[0110] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0111] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0112] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit with a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit with a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0113] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A control method for a LED light source multi-light point projection module for environment rendering, characterized in that: include, Collect user needs and user browsing data through the app, and connect the app with the light spot projector; Construct a control model, input user requirements into the control model to obtain a control strategy, control the light spot projector according to the control strategy, and record the control strategy as a user personalized solution; Predict user preferences based on user personalization plans and user browsing data, and push them to users through the app; Users can remotely modify user needs to control the light spot projector through the app; The collecting of user needs and user browsing data through the app and connecting the app with the light spot projector refers to connecting the app with the light spot projector, collecting user demand data through the app, including light spot brightness, color, flickering frequency, quantity, distribution and projection distance, cleaning and filtering the user demand data after the app collects it, and simultaneously collecting the user's browsing data in the app; The construction of the control model and inputting the user requirements into the control model to obtain the control strategy include: The initial state data of the light spot projector is collected and preprocessed through the app, and feature engineering is used to extract data features of the initial state data and user demand data; The initial state data feature of the light spot projector is used as the initial state vector , and use the user demand data features as the ideal state vector of the light spot projector , based on the initial state vector and the ideal state vector Build the topology structure and convert the initial state vector and the ideal state vector The data types are used as layers, the data features in each data are used as nodes in the layers, the associations between nodes are used as edges, the node weights are set and the Pearson correlation coefficients between nodes are calculated as the node edge weights; Define the local state tracking error in the topology : ;in is the initial state vector The eigenvalue of the i-th node in the topological structure, is the ideal state vector The eigenvalue of the i-th node in the topological structure, n is the total number of nodes in the topological network; Initialize the adaptive control gain parameters and construct the adaptive control law to generate the control input : ;in , as well as Adjust , as well as The control gain parameters of Solve the first positive definite matrix P through the Riccati equation and define the diagonal matrix as the second positive definite matrix , construct the Lyapunov function , define the optimization constraints as: ； ； ； ;in , as well as is the updated control gain parameter, T is the transposition operation; Solve the Lyapunov function by optimizing the constraints to obtain the updated control gain parameters and bring them into the adaptive control law to obtain the updated control input ; The consensus weight between nodes is calculated based on the node weight in the topology: ; Where C is the consensus weight, m ​​is the number of topological structure levels, and n is the number of nodes in the topological structure. is the smoothing parameter, is the weight of the i-th node in the k-th level, is the weight of the jth node at the kth level; Control input via consensus weights and updates The final control input Q is obtained: ;in is the attenuation factor; Select the RBF kernel function to construct the control model, set the initial kernel function parameters and regularization parameters, and define the control model input as the final control input Q, and the model output as the state vector of the light spot projector; Define the model optimization objective function, use the model training set input to control the model for model iterative training, and label the model training set data; In the iterative process, a parameter combination is formed for each kernel function parameter and regularization parameter within the defined parameter range, the Lagrange multiplier corresponding to each parameter combination is calculated, and the model optimization objective function value is calculated, and the parameter combination that minimizes the model optimization objective function value is selected as the optimal parameter combination, and its corresponding Lagrange multiplier is used as the model input; The final control input Q is input into the control model to obtain the state vector of the light spot projector , define the objective function as: ; Solve the objective function value and set the comparison threshold. When the objective function value is greater than the comparison threshold, the state vector of the light spot projector is As the initial state vector of the light spot projector, it is re-optimized until the objective function value is less than or equal to the comparison threshold, and then the optimization is stopped and the corresponding final control input Q is output; The light spot projector control strategy is formed according to the corresponding final control input Q.

2. The control method of the LED light source multi-light point projection module for environment rendering according to claim 1, characterized in that: The controlling the light spot projector according to the control strategy and recording the control strategy as a user personalized plan means that the app obtains the control strategy of the light spot projector through the control model and controls the light spot projector to perform light spot projection according to the control strategy, the app records the control strategy and the light spot projection map, and forms a user personalized plan based on the user data and stores it in the user's personal database, allowing the user to upload the user's personalized plan to the app cloud space for sharing the personalized plan, the app cloud space will display the shared personalized plan, and allow the user to browse and download the personalized plan from the app cloud space and switch to downloading the personalized plan with one click.

3. The control method of the LED light source multi-light point projection module for environment rendering according to claim 2, characterized in that: The prediction of user preferences based on user personality schemes and user browsing data and the push notification to users through the app include: Construct a convolutional neural network, which includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer; Set the input layer input format to the user's personalized plan and user browsing data format, and connect it to the convolution layer; The convolution layer is designed as a complex layer structure. Each convolution layer uses the ReLU function as the activation function, sets 32 filters, the filter size is 3*3, and the step size is 1; The pooling layer is connected after the convolutional layer to perform maximum pooling on the convolutional layer output, setting the maximum pooling operation size to 2*2 and the step size to 2; The fully connected layer uses the sigmoid function as the activation function to process the output of the pooling layer and outputs it through the output layer; Use the neural network training set to train the convolutional neural network, annotate the data in the neural network training set, set the mean square error as the loss function, use the Adam optimizer to iteratively optimize the convolutional neural network parameters, calculate the error of the convolutional neural network in each iteration, and stop the iteration to obtain the optimal convolutional neural network parameters when the error no longer decreases significantly during continuous iterations; The user's personalized plan and user browsing data are input into the convolutional neural network to obtain the user's personalized scene push, and the correlation between the user's personalized scene push and the user's personalized plan is calculated. A personalized scene push list is formed from high to low according to the correlation and pushed to the user through the app.

4. The control method of the LED light source multi-light point projection module for environment rendering according to claim 3, characterized in that: The user remotely modifies the user needs to control the light spot projector through the app means that after the user modifies the user needs on the app, the app automatically generates the modified user need data, and inputs it into the control model to obtain the control strategy to control the light spot projector. The app simultaneously collects the user voice data, converts the user voice data into text data based on voice recognition technology, and extracts the user needs from the text data through natural language processing technology to automatically generate the modified user need data.

5. An LED light source multi-light point projection module for environment rendering, characterized in that: It includes the control method as described in any one of claims 1 to 4, and also includes a light spot projector and a processing module, wherein the light spot projector includes a light source, a reflector, a convex mirror and a scattering mirror, and the light source emits light which is reflected by the reflector, passes through the convex mirror, and is then scattered into a light spot by the scattering mirror.

6. The LED light source multi-light point projection module for environment rendering according to claim 5, characterized in that: The processing module comprises: Data collection module, used to collect and process user demand data and user browsing data; The device control module is used to build a control model and obtain a control strategy to control the light spot projector according to the user demand data analysis; Personalized push module, used to store user personalized plans and predict user personalized scenarios based on user personalized plans and user browsing data and push them to users through the app; The remote control module is used for users to remotely modify user needs through the app to control the light spot projector and collect user voice to recognize user needs.

Images