Self-learning Internet of Things induction lamp and control method thereof

Abstract

The invention discloses a self-learning Internet of Things induction lamp and a control method thereof, and belongs to the technical field of intelligent illumination. The device comprises a power supply, sensors, a processing module, a communication module and a lighting module, data are collected through 5.8 G radar, infrared and illuminance sensors, automatic networking of lamps, advancing direction recognition and multi-sensor fusion are achieved by utilizing DBSCAN clustering, an HMM model and a Kalman filtering algorithm, and a dynamic lighting strategy is generated. The control method comprises the steps of data acquisition, self-learning networking, communication interaction and dynamic control, supports Bluetooth Mesh and WiFi, can adaptively adjust brightness according to ambient light intensity, illuminates as required in an induction mode, and is efficient and energy-saving in a sleep mode. The system solves the problems of high energy consumption, complex debugging and the like of traditional illumination, has the advantages of debugging-free rapid deployment, more than 70% of energy saving rate, high distributed reliability and the like, is suitable for scenes such as garages and corridors, and realizes intelligent upgrading of the illumination system.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent lighting, and particularly relates to a self-learning Internet of Things induction lamp and a control method thereof. Background Art

[0002] Currently, traditional commercial lighting systems have the following significant defects in scenarios such as garages, corridors, and underground spaces:

[0003] Low energy consumption efficiency: Traditional LED lamps are usually in a constant-on state. For example, an ordinary 18W LED lamp consumes 0.43 degrees of electricity per day on average, resulting in serious energy waste. Even with single-lamp induction technology, only the single lamp lights up when someone passes by, with limited energy-saving effect and lighting blind spots.

[0004] Complex debugging and deployment: Existing self-organizing network induction lamps rely on professionals to plan the positions of lamps on-site and manually configure linkage logic. The debugging cost accounts for more than 30% of the total project cost, and the wiring is complex, making it difficult to meet the needs of the renovation market.

[0005] Poor lighting experience: In the single-lamp induction mode, the lamp only lights up passively when the target approaches, resulting in a dark field of vision ahead and potential safety hazards; at the same time, the traditional system lacks the ability to adapt to ambient light. It still maintains high brightness when the daylight is sufficient during the day, further exacerbating energy consumption.

[0006] Insufficient system reliability: Under the centralized control architecture, a single-point failure easily leads to local lighting failure, and after-sales service relies on professionals to visit on-site, with high operation and maintenance costs.

[0007] In response to the above problems, although the existing technology has proposed a wireless self-organizing network solution, it has not solved the core contradiction of "commissioning-free self-organization" and "dynamic energy-saving control". There is an urgent need for an intelligent lighting system with self-learning ability and high reliability. Summary of the Invention

[0008] The purpose of the present invention is to provide a self-learning Internet of Things induction lamp and a control method thereof, which solve the problems of low energy consumption efficiency, complex debugging and deployment, poor lighting experience, and insufficient system reliability.

[0009] To achieve the above purpose, the present invention provides a self-learning Internet of Things induction lamp, including a power supply module, the power supply module is electrically connected to a sensor module, a processing module, and a lighting module respectively, the sensor module is electrically connected to the processing module, and the processing module is electrically connected to a communication module;

[0010] The sensor module includes a 5.8G radar sensor, an infrared sensor, and a light intensity sensor;

[0011] The processing module includes a TLSRL825X main control chip, a Flash storage unit, and a RAM storage unit;

[0012] The communication module includes a Bluetooth Mesh module and a WiFi module;

[0013] The power module includes an AC-DC drive circuit (input voltage 85 - 265V, output DC5 - 12V, conversion efficiency ≥90%, powering the sensor, processing, and communication modules) and a constant current drive circuit (based on PWM dimming technology, output current accuracy ±3%, supporting 0 - 100% stepless dimming, controlling the power of the LED light source);

[0014] The lighting module includes an LED light source (18W / 36W optional, luminous efficacy ≥100lm / W, color rendering index Ra≥80, supporting 0 - 100% brightness adjustment) and an optical lens (beam angle 60° / 120° optional, achieving uniform lighting or enhanced lighting in key areas).

[0015] Preferably, the 5.8G radar sensor is connected to the processing module through an I 2 C interface for transmitting target position and speed information;

[0016] The infrared sensor uses the GPIO interface to send digital signals to the processing module;

[0017] The illuminance sensor transmits the analog voltage signal to the processing module through an I 2 C interface.

[0018] Preferably, the TLSRL825X main control chip is respectively connected to the Flash and the RAM through the SPI interface for data storage and reading;

[0019] Both the Bluetooth Mesh module and the WiFi module are connected to the TLSRL825X main control chip. The Bluetooth Mesh module is used for communication between local lamps, and the WiFi module is used for remote control.

[0020] Preferably, the AC-DC drive circuit is connected to an AC85 - 265V power supply, outputs DC5 - 12V to power the processing module, the sensor module, and the communication module; outputs DC36 - 108V to power the lighting module; the constant current drive circuit is connected to the LED light source for adjusting the LED current according to the instructions of the processing module;

[0021] The optical lens is installed in front of the LED light source for adjusting the light propagation direction and range.

[0022] The present invention also provides a control method for a self-learning Internet of Things induction lamp, including the following steps:

[0023] S1. Data acquisition: Real-time collect environmental data through a 5.8G radar sensor, an infrared sensor, and an illuminance sensor to construct a multi-dimensional input signal;

[0024] S2, Self-learning Networking and Policy Generation: Based on the data collected in step S1, divide the lamp groups through DBSCAN clustering, identify the traveling direction using the HMM model, and fuse multi-sensor signals with Kalman filtering to generate a dynamic lighting policy;

[0025] S3, Communication Interaction: Achieve local policy sharing and fault self-healing through Bluetooth Mesh, support remote parameter configuration through WiFi, and ensure the coordination of multiple devices;

[0026] S4, Dynamic Control: Control the brightness and power consumption of the LED light source according to the policy generated in step S2, and achieve dynamic lighting and adaptive illuminance in different scenarios.

[0027] Preferably, in step S1, the illuminance sensor converts the ambient illuminance L into a corresponding analog voltage signal V lux , which is convenient for the processing module to perform digital processing;

[0028] The output formula of the illuminance sensor is as follows:

[0029] V lux = k·log(L + 1)

[0030] where V lux is the analog voltage signal output by the illuminance sensor, k is the sensitivity coefficient of the sensor, and L is the ambient illuminance.

[0031] Preferably, in step S2, the specific steps of self-learning networking and policy generation are as follows:

[0032] Use the DBSCAN clustering algorithm to group the lamps. By calculating the spatial distance between the lamps, determine which lamps are closely adjacent in space, and thus divide them into different groups. Set different distance thresholds and core point number thresholds according to the specific application scenario, and the divided groups can meet the actual lighting requirements;

[0033] The expression of DBSCAN clustering is:

[0034]

[0035] Cluster(l j ) represents the clustering set formed with the lamp l j as the core point, and l j and l k represent the position coordinates of different lamps respectively; d(l j ,l k ) is the distance between the lamp l j and l kThe Euclidean distance between them, and its calculation formula is (x j , y j ) and (x k , y k ) are the two-dimensional plane coordinates of the lamps l j and l k respectively; ε is the distance threshold used to define the adjacent range between the lamps; N ε (l j ) represents the set of lamps whose distance from the lamp l j is less than or equal to ε. When MinPts, l j can be used as a core point to form a cluster;

[0036] Then, use the HMM model to identify the moving direction of the target object. The HMM model is based on historical trajectory data to statistically obtain the state transition matrix A and the observation probability matrix B, and find the most likely moving direction through the real-time input target object speed vector by using the Viterbi algorithm;

[0037] The expression of the elements of the HMM model state transition matrix is:

[0038] A ij = P(S t+1 = j|S t = i);

[0039] Among them, A ij is the element in the i-th row and j-th column of the state transition matrix A; S t represents the state at time t, and the state space = {going straight, turning left, turning right, stationary}; P(S t-1 = j|S t = i) represents the probability of transitioning to state j at time (t + 1) under the condition of being in state i at time t;

[0040] Finally, use the Kalman filter algorithm to fuse the data from the radar sensor and the infrared sensor, filter out the noise interference, and obtain more accurate target state information;

[0041] The Kalman filter state transition equation is used to predict the state of the target object at the next moment. Considering the motion law of the target object and the possible noise interference during the process, the state X t at the current moment is mapped to the state X t+1 at the next moment through the state transition matrix F, providing a basis for subsequent state estimation.

[0042] The Kalman filter state transition equation is as follows:

[0043] X t+1 = FXt +W t ;

[0044] Among them, X t+1 and X t are the target state vectors at times (t + 1) and t respectively, x, y are the position coordinates of the target object, is the speed of the target object in the x and y directions; F is the state transition matrix, which is used to describe the variation law of the target object's state over time; W t is the process noise vector, which represents the state prediction error caused by various uncertain factors, and W t obeys a Gaussian distribution with zero mean.

[0045] The Kalman filter observation equation relates the true state X t of the target object to the observation value Z t of the sensor. Considering the noise interference in the sensor measurement process, the target state vector X t is converted into the observation vector Z t through the observation matrix H, providing observation information for subsequent state estimation.

[0046] The Kalman filter observation equation is as follows:

[0047] Z t = HX t + V t ;

[0048] Among them, Z t is the sensor observation vector at time t; H is the observation matrix, which converts the target state vector X t into a form that the sensor can observe; V t is the observation noise vector, which represents the deviation between the observation value and the true value caused by the measurement error of the sensor itself and external environmental interference factors, and V t obeys a Gaussian distribution with zero mean.

[0049] Preferably, in step S3, the specific steps of communication interaction are as follows:

[0050] Through the Bluetooth Mesh network, each lamp encodes its own state information (such as current brightness, whether it is working properly, etc.) and the control strategy generated according to step S2 (such as lighting range, delay time, etc.) into Bluetooth Mesh data packets in a specific format, and broadcasts them to the surrounding adjacent lamps at a certain time period.

[0051] After receiving these data packets, the adjacent lamps will parse them and update the locally stored group table and control parameters, thereby realizing the sharing of local strategies.

[0052] If a certain luminaire does not receive the broadcast signal of an adjacent luminaire within a period of time, it is determined that the adjacent luminaire may malfunction. At this time, it will be removed from the current linkage relationship and trigger the network reorganization mechanism to ensure the stability and reliability of the entire lighting network.

[0053] At the same time, the system is connected to the cloud server through the WiFi module, and users can send parameter configuration instructions to the cloud server through the mobile phone APP or other remote control terminals.

[0054] The cloud server forwards these instructions to the corresponding luminaires, and the luminaires update the parameters after receiving the instructions, so as to realize the remote parameter configuration function and ensure the coordination and consistency among multiple devices.

[0055] Among them, the Bluetooth Mesh data packet format is:

[0056] Packet=[SourrceID,TargetID,Command,Data];

[0057] Packet represents a complete Bluetooth Mesh data packet; SourceID is the unique identifier of the luminaire that sends the data packet, used to identify the source of the data packet; TargetID is the unique identifier of the luminaire that receives the data packet. If it is a broadcast packet, TargetID can be set to a specific broadcast identifier, used to indicate that the data packet needs to be sent to all luminaires or luminaires within a specific range in the network; Command represents the type of instruction carried by the data packet, such as updating the group table, setting the brightness, querying the status, etc.; Data is the specific data content, which varies according to the different Command. For example, when Command is to set the brightness, Data is the brightness value to be set.

[0058] Preferably, in step S4, the steps of dynamic lighting are as follows:

[0059] According to the dynamic lighting strategy generated in step S2, the brightness and power consumption of the LED light source are accurately controlled to meet the requirements of different scenarios. When specific trigger conditions are met, that is, the distance d between the target object and the luminaire dist is less than or equal to the set distance threshold D th 、the moving speed v of the target object speed is greater than or equal to the set speed threshold V min and the infrared sensor detects the presence of the target object (S IR= 1), the system will adjust the brightness of the current lamp and the two lamps in front to the maximum brightness, and at the same time adjust the brightness of one lamp behind to 50% of the maximum brightness, forming a reasonable lighting area to provide good lighting conditions for the movement of the target object. If no new target object activity is detected within the set delay time T delay the system will control the lamp to enter the sleep mode, reduce the brightness to a lower level, defaulting to 10% of the maximum brightness, to reduce power consumption and achieve the purpose of energy conservation;

[0060] The expression for the trigger condition of the induction mode is:

[0061] Trigger = (d dist ≤ D th ) ∧ (v speed ≥ V min ) ^ (S IR = 1);

[0062] Among them, Trigger is a boolean variable, which takes the value of true when the condition is met, otherwise it takes the value of false; d dist is the distance between the target object and the lamp; D th is the distance threshold, used to define the distance range between the target object and the lamp; v speed is the moving speed of the target object; V min is the speed threshold, used to screen out the target objects with actual movement requirements; S IR is the output signal of the infrared sensor. When a target object is detected, S IR = 1, and when no target object is detected, S IR = 0.

[0063] Preferably, in step S4, the steps of light adaption are as follows:

[0064] The present invention will also dynamically adjust the brightness of the lamp according to the ambient light intensity L collected by the light intensity sensor, avoiding over-illumination in situations with sufficient light such as during the day, and further improving the energy utilization efficiency;

[0065] The expression for light adaption is as follows:

[0066]

[0067] Among them, B real is the actual brightness of the lamp, B max is the maximum brightness of the lamp, and L is the ambient light intensity;

[0068] When the ambient light intensity L is less than 200 lux, the actual brightness of the lamp is calculated according to the formula Perform calculations; when the ambient light intensity L is greater than or equal to 200 lux, the actual brightness of the lamp is fixed at 10% of the maximum brightness.

[0069] Therefore, the self - learning Internet of Things induction lamp and its control method with the above - mentioned structure of the present invention have the following beneficial effects:

[0070] (1) Based on the DBSCAN clustering algorithm, after the lamp is powered on, it automatically collects ambient data, divides the linkage group and the intersection group without manual intervention, and greatly improves the deployment efficiency; it provides on - demand lighting (100% brightness) in the induction mode, reduces power consumption in the sleep mode, combines the illuminance sensor and the adaptive formula, automatically reduces the brightness during the day, avoids over - illumination, and balances energy conservation and visibility.

[0071] (2) The present invention adopts the Bluetooth Mesh decentralized architecture, and the failure of a single lamp does not affect the operation of the system, improving the overall stability of the system; the 5.8G radar and the infrared sensor cooperate for detection, combined with the Kalman filtering algorithm, effectively reduce false triggers; based on the HMM model to identify the moving direction, achieve the effect of early lighting, and improve the user experience and safety.

[0072] The technical solution of the present invention will be further described in detail below through the drawings and embodiments. Brief Description of the Drawings

[0073] Figure 1 is the overall schematic diagram of a self - learning Internet of Things induction lamp of the present invention;

[0074] Figure 2 is the schematic diagram of the processing module of a self - learning Internet of Things induction lamp of the present invention;

[0075] Figure 3 is the schematic diagram of the sensor module of a self - learning Internet of Things induction lamp of the present invention;

[0076] Figure 4 is the schematic diagram of the communication module of a self - learning Internet of Things induction lamp of the present invention;

[0077] Figure 5 is the schematic diagram of the lighting module of a self - learning Internet of Things induction lamp of the present invention;

[0078] Figure 6 is the schematic diagram of the power supply module of a self - learning Internet of Things induction lamp of the present invention;

[0079] Figure 7 is the schematic diagram of the flow of the control method of a self - learning Internet of Things induction lamp of the present invention. Detailed Embodiments

[0080] The technical solution of the present invention will be further described below through the drawings and embodiments.

[0081] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings as understood by those of ordinary skill in the art to which the present invention pertains. The "first", "second" and similar terms used in the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. The terms such as "comprising" or "including" mean that the elements or items appearing before this term cover the elements or items listed after this term and their equivalents, without excluding other elements or items. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0082] Embodiment

[0083] As Figures 1-6 shown, the present invention provides a self-learning Internet of Things induction lamp, including a power supply module, the power supply module is electrically connected to a sensor module, a processing module and a lighting module respectively, the sensor module is electrically connected to the processing module, and the processing module is electrically connected to a communication module;

[0084] The sensor module includes a 5.8G radar sensor, an infrared sensor and a light intensity sensor;

[0085] The processing module includes a TLSRL825X main control chip, a Flash storage unit and a RAM storage unit;

[0086] The communication module includes a Bluetooth Mesh module and a WiFi module;

[0087] The power supply module includes an AC-DC drive circuit (input voltage 85 - 265V, output DC5 - 12V, conversion efficiency ≥ 90%, powering the sensor, processing, and communication modules) and a constant current drive circuit (based on PWM dimming technology, output current accuracy ±3%, supporting 0 - 100% stepless dimming, controlling the power of the LED light source);

[0088] The lighting module includes an LED light source (18W / 36W optional, light efficiency ≥ 100lm / W, color rendering index Ra ≥ 80, supporting 0 - 100% brightness adjustment) and an optical lens (beam angle 60° / 120° optional, achieving uniform lighting or enhanced lighting in key areas).

[0089] The 5.8G radar sensor is connected to the processing module through an I 2 C interface for transmitting target position and speed information;

[0090] The infrared sensor uses the GPIO interface to send digital signals to the processing module;

[0091] The illuminance sensor transmits the analog voltage signal to the processing module through the I 2 C interface.

[0092] The TLSRL825X main control chip is respectively connected to the Flash and the RAM through the SPI interface for data storage and reading;

[0093] Both the Bluetooth Mesh module and the WiFi module are connected to the TLSRL825X main control chip. The Bluetooth Mesh module is used for communication between local lamps, and the WiFi module is used for remote control.

[0094] The AC-DC drive circuit is connected to the AC85-265V power supply and outputs DC5-12V power supply; the constant current drive circuit is connected to the LED light source and is used to adjust the LED current according to the instructions of the processing module;

[0095] The optical lens is installed in front of the LED light source and is used to adjust the light propagation direction and range.

[0096] As Figure 7 shown, the present invention also provides a control method for a self-learning Internet of Things induction lamp, including the following steps:

[0097] S1. Data acquisition: Real-time collect environmental data through the 5.8G radar sensor, infrared sensor, and illuminance sensor to construct a multi-dimensional input signal;

[0098] The 5.8G radar emits a frequency-modulated continuous wave at an interval of 0.1 second, and after receiving the reflected signal, calculates the target coordinates (x, y, z) and speed through the I 2 C interface and transmits it to the processing module;

[0099] The infrared sensor continuously monitors moving objects. When detecting a change in human / vehicle thermal radiation, it outputs a high-level signal S IR =1, otherwise 0, and inputs it to the processing module through the GPIO interface.

[0100] The illuminance sensor converts the environmental illuminance L into a corresponding analog voltage signal V lux for digital processing by the processing module;

[0101] The output formula of the illuminance sensor is as follows:

[0102] V lux =k·log(L + 1)

[0103] where V luxis the analog voltage signal output by the illuminance sensor, k is the sensitivity coefficient of the sensor, and L is the ambient illuminance.

[0104] S2. Self-learning networking and strategy generation: Based on the data collected in step S1, divide the lamp groups through DBSCAN clustering, identify the traveling direction by the HMM model, and combine the Kalman filter to fuse multi-sensor signals to generate a dynamic lighting strategy.

[0105] The specific steps of self-learning networking and strategy generation are as follows:

[0106] Use the DBSCAN clustering algorithm to group the lamps. By calculating the spatial distance between the lamps, determine which lamps are closely adjacent in space, and thus divide them into different groups. Set different distance thresholds and core point number thresholds according to the specific application scenario, and the divided groups can meet the actual lighting requirements.

[0107] The expression of DBSCAN clustering is:

[0108] Cluster(l j ) = {l k |d(l j , l k ) ≤ ε, | N ε (l j ) | ≥ MinPts};

[0109] Cluster(l j ) represents the clustering set formed with the lamp l j as the core point, and l j and l k represent the position coordinates of different lamps respectively; d(l j , l k ) is the Euclidean distance between the lamps l j and l k , and its calculation formula is (x j , y j ) and (x k , y k ) are the two-dimensional plane coordinates of the lamps l j and l k respectively; ε is the distance threshold used to define the adjacent range between the lamps; N ε (l j ) represents the set of lamps whose distance from the lamp l j is less than or equal to ε, and when MinPts, l j can be used as the core point to form a cluster.

[0110] Then, use the HMM model to identify the moving direction of the target object. The HMM model is based on historical trajectory data to statistically obtain the state transition matrix A and the observation probability matrix B, and through the real-time input target object velocity vector Use the Viterbi algorithm to find the most likely moving direction;

[0111] The expression of the elements of the HMM model state transition matrix is:

[0112] A ij = P(S t+1 = j|S t = i);

[0113] Among them, A ij is the element in the i-th row and j-th column of the state transition matrix A; S t represents the state at time t, and the state space = {going straight, turning left, turning right, stationary}; P(S t-1 = j|S t = i) represents the probability of transitioning to state j at time (t + 1) under the condition of being in state i at time t;

[0114] Finally, use the Kalman filter algorithm to fuse the data from the radar sensor and the infrared sensor, filter out the noise interference, and obtain more accurate target state information;

[0115] The Kalman filter state transition equation is used to predict the state of the target object at the next moment. Considering the motion law of the target object and the possible noise interference in the process, the state at the current moment X t is mapped to the state X t+1 at the next moment through the state transition matrix F, providing a basis for subsequent state estimation.

[0116] The Kalman filter state transition equation is as follows:

[0117] X t+1 = FX t + W t ;

[0118] Among them, X t+1 and X t are the target state vectors at time (t + 1) and t respectively, x, y are the position coordinates of the target object, is the velocity of the target object in the x and y directions; F is the state transition matrix, used to describe the change law of the target object state over time; W t is the process noise vector, representing the state prediction error caused by various uncertain factors, and W t obeys a Gaussian distribution with zero mean.

[0119] The Kalman filter observation equation relates the true state X of the target object t to the observed value Z of the sensor t . Considering the noise interference in the sensor measurement process, the target state vector X is converted into the observed vector Z through the observation matrix H t to provide observation information for subsequent state estimation t .

[0120] The Kalman filter observation equation is as follows

[0121] Z t = HX t + V t ;

[0122] where Z t is the sensor observation vector at time t; H is the observation matrix that converts the target state vector X t into a form observable by the sensor; V t is the observation noise vector, representing the deviation between the observed value and the true value caused by the measurement error of the sensor itself and external environmental interference factors, and V t follows a Gaussian distribution with zero mean

[0123] S3. Communication and interaction: Local policy sharing and fault self-healing are achieved through Bluetooth Mesh, remote parameter configuration is supported through WiFi, and multi-device coordination is ensured

[0124] The specific steps of communication and interaction are as follows

[0125] Through the Bluetooth Mesh network, each lamp encodes its own status information (such as current brightness, whether it is working properly, etc.) and the control strategy generated according to step S2 (such as lighting range, delay time, etc.) into Bluetooth Mesh data packets in a specific format and broadcasts them to the surrounding adjacent lamps at a certain time interval

[0126] After receiving these data packets, the adjacent lamps will parse them and update the locally stored group table and control parameters, thereby realizing local policy sharing

[0127] If a lamp does not receive the broadcast signal of an adjacent lamp within a certain period of time, it will determine that the adjacent lamp may have a fault. At this time, it will remove it from the current linkage relationship and trigger the network reorganization mechanism to ensure the stability and reliability of the entire lighting network

[0128] At the same time, the system connects to the cloud server through the WiFi module, and users can send parameter configuration instructions to the cloud server through the mobile APP or other remote control terminals

[0129] The cloud server forwards these instructions to the corresponding lamps. After receiving the instructions, the lamps update their parameters, thus realizing the remote parameter configuration function and ensuring the coordination and consistency among multiple devices.

[0130] Among them, the Bluetooth Mesh data packet format is:

[0131] Packet=[SourrceID, TargetID, Command, Data];

[0132] Packet represents a complete Bluetooth Mesh data packet; SourceID is the unique identifier of the lamp that sends this data packet, used to identify the source of the data packet; TargetID is the unique identifier of the lamp that receives this data packet. If it is a broadcast packet, TargetID can be set to a specific broadcast identifier, used to indicate that this data packet needs to be sent to all lamps or lamps within a specific range in the network; Command represents the type of instruction carried by the data packet, such as updating the group table, setting the brightness, querying the status, etc.; Data is the specific data content, which varies according to different Commands. For example, when Command is setting the brightness, Data is the brightness value to be set.

[0133] S4. Dynamic control: According to the strategy generated in step S2, control the brightness and power consumption of the LED light source to achieve dynamic lighting and adaptive illuminance in different scenarios.

[0134] The steps of dynamic lighting are as follows:

[0135] According to the dynamic lighting strategy generated in step S2, precisely control the brightness and power consumption of the LED light source to adapt to different scenario requirements. When specific trigger conditions are met, that is, the distance d dist between the target object and the lamp is less than or equal to the set distance threshold D th , the moving speed v speed of the target object is greater than or equal to the set speed threshold V min and the infrared sensor detects the presence of the target object (S IR = 1), the system will adjust the brightness of the current lamp and the two lamps in front to the maximum brightness, and at the same time adjust the brightness of the lamp behind to 50% of the maximum brightness, forming a reasonable lighting area to provide good lighting conditions for the movement of the target object. If no new target object activities are detected within the set delay time T delay , the system will control the lamp to enter the sleep mode, reduce the brightness to a lower level, by default 10% of the maximum brightness, to reduce power consumption and achieve the purpose of energy conservation;

[0136] The expression of the induction mode trigger condition is:

[0137] Trigger = (d dist ≤D th ) ^ (v speed ≥V min ) ∧ (S IR == 1);

[0138] Where Trigger is a boolean variable, which takes the value true when the condition is met, and false otherwise; d dist is the distance between the target object and the lamp; D th is the distance threshold for defining the distance range between the target object and the lamp; v speed is the moving speed of the target object; V min is the speed threshold for screening out target objects with actual movement requirements; S IR is the output signal of the infrared sensor. When a target object is detected, S IR == 1, and when no target object is detected, S IR == 0.

[0139] The steps of light intensity adaptation are as follows:

[0140] The present invention also dynamically adjusts the brightness of the lamp according to the ambient light intensity L collected by the light intensity sensor, avoiding over-illumination in well-lit situations such as during the day, and further improving the energy utilization efficiency;

[0141] The expression of light intensity adaptation is as follows:

[0142]

[0143] Where B real is the actual brightness of the lamp, B max is the maximum brightness of the lamp, and L is the ambient light intensity;

[0144] When the ambient light intensity L is less than 200 lux, the actual brightness of the lamp is calculated according to the formula When the ambient light intensity L is greater than or equal to 200 lux, the actual brightness of the lamp is fixed at 10% of the maximum brightness.

[0145] Taking the deployment of the intelligent lighting system in an underground parking lot as an example, it is as follows:

[0146] (I) Hardware installation and system initialization

[0147] Lamp configuration: Select an 18W radar-sensing LED lamp (including a 5.8G radar sensor, an infrared sensor, and a light intensity sensor), and set the optical lens beam angle to 120°, covering the parking lot lane and parking space areas.

[0148] Power connection: The AC-DC drive circuit is connected to the 220V mains power supply, and outputs DC12V to supply power to the sensors and the processing module; the constant current drive circuit controls the LED light source and supports 0-100% PWM dimming.

[0149] Communication module: The Bluetooth Mesh module enables the self-organizing network mode, and the WiFi module connects to the cloud server through the existing WiFi network in the parking lot.

[0150] Initial configuration: After each lamp is powered on, it broadcasts its own ID (such as L001, L002) and the initial position (preset based on the installation coordinates) through Bluetooth, and adjacent lamps automatically establish connections to form an initial network.

[0151] (2) Data acquisition and self-learning networking

[0152] (1) Multi-dimensional data acquisition:

[0153] Radar sensor: Collects the position coordinates (x, y) and speed of vehicles / pedestrians at intervals of 0.1 second The detection range is 5-7 meters, and the accuracy is ±0.5 meters.

[0154] Infrared sensor: Outputs the digital signal S in real time IR When a target is detected, a high level S is triggered IR = 1.

[0155] Illuminance sensor: Collects the ambient light intensity L, and converts it into an analog voltage signal through V lux = k·log(L + 1), and inputs it into the processing module for A / D conversion.

[0156] (2) Self-learning networking

[0157] DBSCAN clustering and grouping:

[0158] Threshold setting: In the lane area, ε = 8 meters, MinPts = 3; in the intersection area, ε = 12 meters, MinPts = 5.

[0159] Grouping process: The processing module calculates the Euclidean distance between lamps Lamps that meet the conditions are divided into a "linkage group" (3 adjacent lane lamps) or an "intersection group" (5 multi-directional intersection lamps), and a group logic table is generated and stored in the Flash unit.

[0160] Recognition of the traveling direction:

[0161] The state space of the HMM model is defined as {going straight, turning left, turning right, stationary}. By statistically analyzing the vehicle traveling data during a 3-day learning period, a state transition matrix A (such as the probability of going straight → turning left is 0.1, and the probability of going straight → going straight is 0.8) and an observation probability matrix B are generated, and the speed vector is input in real time Identify the current direction.

[0162] (III) Communication Interaction and Policy Synchronization

[0163] Local Mesh Networking: Each lamp broadcasts a status packet (taking the current brightness and group ID as an example) every 10 seconds. Adjacent lamps update the local group table after receiving it. If lamp L003 fails, adjacent lamps L002 and L004 automatically remove the linkage relationship of L003 after not receiving a signal for 30 seconds, triggering network reorganization to ensure no blind spots in the lighting area.

[0164] Remote Parameter Configuration: The user sets the "peak mode" through the mobile APP: the lighting range is extended to the 5 lamps in front, the delay time is set to 20 seconds, and the sleep brightness is adjusted to 15%. The instruction is transmitted to the cloud through the WiFi module and then sent to the target lamp, and the parameter update is completed within 5 seconds.

[0165] (IV) Dynamic Control and Energy-saving Execution

[0166] Induction Mode Trigger: When the distance d of the vehicle from the lamp dist ≤ 3m, the speed v speed ≥ 0.2m / s and S IR = 1, the brightness of the current lamp (L001) and the 2 lamps in front (L002, L003) is adjusted to 100%, and the 1 lamp behind (L000) is adjusted to 50%, forming a 15-meter lighting range for 10 seconds.

[0167] Sleep Mode and Light Adaptation: If there is no new target within 10 seconds, the lamp enters the sleep mode and the brightness drops to 10%. When the ambient light intensity L is greater than or equal to 200 lux (daytime), the 10% brightness is automatically maintained; if the ambient light intensity L is less than 200 lux (night), perform Dynamic dimming.

[0168] Finally, it should be noted that: The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions of the present invention or make equivalent replacements, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A self-learning IoT sensor lamp, characterized in that: The system comprises a power supply module, wherein the power supply module is electrically connected to the sensor module, the processing module and the lighting module respectively, the sensor module is electrically connected to the processing module, and the processing module is electrically connected to the communication module; The sensor module includes a 5.8G radar sensor, an infrared sensor, and an illumination sensor; The processing module includes a TLSRL825X main control chip, a Flash storage unit and a RAM storage unit; The communication module includes a Bluetooth Mesh module and a WiFi module; The power module includes an AC-DC drive circuit and a constant current drive circuit; The lighting module includes an LED light source and an optical lens.

2. The self-learning IoT sensor lamp according to claim 1, characterized in that: 5.8G radar sensor through I 2 The C interface is connected to the processing module and is used to transmit the target position and speed information; The infrared sensor uses the GPIO interface to send digital signals to the processing module; The light sensor is connected via I 2 C interface, transmits the analog voltage signal to the processing module.

3. The self-learning Internet of Things sensor lamp according to claim 1, characterized in that: The TLSRL825X main control chip is connected to the Flash and RAM through the SPI interface for data storage and reading; Both the Bluetooth Mesh module and the WiFi module are connected to the TLSRL825X main control chip. The Bluetooth Mesh module is used for communication between local lamps, and the WiFi module is used to achieve remote control.

4. The self-learning Internet of Things sensor lamp according to claim 1, characterized in that: The AC-DC drive circuit is connected to an AC85-265V power supply, and outputs DC5-12V to power the processing module, sensor module, and communication module; it outputs DC36-108V to power the lighting module; the constant current drive circuit is connected to the LED light source, and is used to adjust the LED current according to the instructions of the processing module; The optical lens is installed in front of the LED light source to adjust the direction and range of light propagation.

5. A control method for a self-learning Internet of Things sensor lamp, applied to a self-learning Internet of Things sensor lamp according to any one of claims 1 to 4, characterized in that: The following steps are involved: S1. Data collection: Collect environmental data in real time through 5.8G radar sensors, infrared sensors and light intensity sensors to build multi-dimensional input signals; S2, self-learning networking and strategy generation: Based on the data collected in step S1, the lamp groups are divided through DBSCAN clustering, the HMM model identifies the direction of travel, and the multi-sensor signals are fused with Kalman filtering to generate dynamic lighting strategies; S3, Communication interaction: local policy sharing and fault self-healing are achieved through Bluetooth Mesh, remote parameter configuration is supported through WiFi, and coordination and consistency among multiple devices are ensured; S4, dynamic control: According to the strategy generated in step S2, the brightness and power consumption of the LED light source are controlled to achieve dynamic lighting and adaptive illumination in different scenarios.

6. The control method of a self-learning Internet of Things induction lamp according to claim 5 is characterized in that: In step S1, the light sensor converts the ambient light intensity L into a corresponding analog voltage signal V lux , which is convenient for the processing module to carry out digital processing; The output formula of the light sensor is as follows: V lux =k log(L+1) Among them, V lux is the analog voltage signal output by the light sensor, k is the sensitivity coefficient of the sensor, and L is the ambient light intensity.

7. The control method of a self-learning Internet of Things sensor lamp according to claim 5 is characterized in that: In step S2, the specific steps of self-learning networking and strategy generation are as follows: Use DBSCAN clustering algorithm to group lamps. By calculating the spatial distance between lamps, determine which lamps are closely adjacent in space, and divide them into different groups. Set different distance thresholds and core point number thresholds according to specific application scenarios. The divided groups can meet the actual lighting needs. The expression of DBSCAN clustering is: Cluster(l j )={l k |d(l j ,L k )≤ε,|N ε (L j )|≥MinPts}; Cluster(l j ) indicates that the lamp is j is the cluster set formed by the core points, l j and l k Represent the position coordinates of different lamps; d(l j ,l k ) is a lamp j and l k The Euclidean distance between them is calculated as (x j ,y j ) and (x k ,y k ) are lamps l j and l k The two-dimensional plane coordinates of ε is the distance threshold, which is used to define the adjacent range between lamps; N ε (l j ) indicates the distance from the lamp l j The set of lamps less than or equal to ε, MinPts, l j Only then can it be used as a core point to form a cluster; The HMM model is then used to identify the direction of travel of the target object. The HMM model is based on historical trajectory data, and statistics are generated by the state transfer matrix A and the observation probability matrix B. The speed vector of the target object is input in real time. Use the Viterbi algorithm to find the most likely direction of travel; The expression of the HMM model state transfer matrix element is: A ij =P(S t+1 =j|S t =i); Among them, A ij is the i-th row and j-th column element in the state transfer matrix A; S t represents the state at time t, state space = {go straight, turn left, turn right, stop}; P(S t-1 =j|S t =i) represents the probability of transitioning to state j at time (t+1) under the condition that the state is in state i at time t; Finally, the Kalman filter algorithm is used to fuse the data from the radar sensor and the infrared sensor to filter out noise interference and obtain more accurate target status information; The Kalman filter state transfer equation is as follows: X t+1 =FX t +W t ; Among them, X t+1 and X t are the target state vectors at time (t+1) and t, respectively. x, y are the position coordinates of the target object, is the speed of the target object in the x and y directions; F is the state transfer matrix, which is used to describe the change of the state of the target object over time; W t is the process noise vector, which represents the state prediction error due to various uncertain factors, and W t It follows a Gaussian distribution with zero mean; The Kalman filter observation equation is as follows: Z t =HX t +V t ; Among them, Z t is the sensor observation vector at time t; H is the observation matrix, which transforms the target state vector X t Converted into a form that can be observed by the sensor; V t is the observation noise vector, which indicates the deviation between the observed value and the true value caused by the measurement error of the sensor itself and the external environmental interference factors, and V t It follows a Gaussian distribution with zero mean.

8. The control method of a self-learning Internet of Things sensor lamp according to claim 5, characterized in that: In step S3, the specific steps of communication interaction are as follows: Through the Bluetooth Mesh network, each lamp encodes its own status information and the control strategy generated according to step S2 into a Bluetooth Mesh data packet, and broadcasts it to the surrounding adjacent lamps at a certain time period; After receiving the above data packets, the adjacent lamps parse them and update the group table and control parameters stored locally to achieve sharing of local strategies; If a lamp does not receive the broadcast signal of a neighboring lamp for a period of time, it is determined that the neighboring lamp may be faulty. At this time, it will be removed from the current linkage relationship and the network reorganization mechanism will be triggered; Then connect to the cloud server through the WiFi module, and the user sends parameter configuration instructions to the cloud server through the mobile phone APP or other remote control terminals; The cloud server forwards the above instructions to the corresponding lamps, and the lamps update the parameters after receiving the instructions, realizing the remote parameter configuration function; Among them, the Bluetooth Mesh data packet format is: Packet=[SourceID, TargetID, Command, Data]; Packet represents a complete Bluetooth Mesh data packet; SourceID is the unique identifier of the lamp that sends the data packet, which is used to identify the source of the data packet; TargetID is the unique identifier of the lamp that receives the data packet, which is used to indicate that the data packet needs to be sent to all lamps in the network or lamps within a specific range; Command represents the command type carried by the data packet; Data is the specific data content.

9. The control method of a self-learning Internet of Things induction lamp according to claim 8 is characterized in that: In step S4, the steps of dynamic lighting are as follows: According to the dynamic lighting strategy generated in step S2, when the induction mode triggering condition is met, the brightness of the current lamp and the two lamps in front are adjusted to the maximum brightness, and the brightness of the lamp in the back is adjusted to 50% of the maximum brightness; The expression of the trigger condition of the induction mode is: Trigger=(d dist ≤D th )∧(v speed ≥V min )^(S IR =1); Trigger is a Boolean variable. When the condition is met, the value is true, otherwise the value is false. dist is the distance between the target and the lamp; D th is the distance threshold, which is used to define the distance range between the target object and the lamp; v speed is the target's moving speed; V min is the speed threshold, which is used to filter out the target objects with actual motion requirements; S IR is the output signal of the infrared sensor. When the target object is detected, S IR =1, when the target object is not detected S IR =0.

10. The control method of a self-learning Internet of Things induction lamp according to claim 5, characterized in that: In step S4, the steps of illumination adaptation are as follows: According to the ambient light intensity L collected by the light sensor, the brightness of the lamp is dynamically adjusted through the light adaptive formula; The expression of light adaptation is as follows: Among them, B real is the actual brightness of the lamp, B max is the maximum brightness of the lamp, and L is the ambient light intensity; When the ambient light intensity L is less than 200 lux, the actual brightness of the lamp is calculated according to the formula Calculation is performed; when the ambient light intensity L is greater than or equal to 200 lux, the actual brightness of the lamp is fixed at 10% of the maximum brightness.

Images