A lighting control method and system in emergency situations

Abstract

The present invention relates to the field of lighting control technology, and discloses a lighting control method and system in an emergency situation, the method comprising: configuring a lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit; monitoring data of the environmental sensor by the control unit; when the data of the environmental sensor indicates the existence of a dangerous situation, the control unit performs a dangerous situation analysis to obtain a danger level; according to the danger level, the control unit generates an emergency lighting plan, wherein the emergency lighting plan includes information of the lighting fixture to be activated; based on the emergency lighting plan, the control unit sends a control instruction to the lighting fixture to be activated; in response to the control instruction, the lighting fixture to be activated switches to an emergency lighting state, the present invention realizes efficient utilization and precise control of lighting resources, and ensures the durability and stability of emergency lighting.

Description

Technical Field

[0001] The present invention relates to the technical field of lighting control, and in particular to a lighting control method and system in emergency situations. Background Art

[0002] With the acceleration of urbanization and the continuous expansion of building scale, the safety of public places and large buildings has received increasing attention. In emergency situations such as fires and earthquakes, effective emergency lighting systems are essential to ensure the safe evacuation of personnel and emergency rescue. Traditional emergency lighting systems usually adopt fixed lighting solutions, lack the ability to respond to environmental changes in real time, and are difficult to adapt to complex and changing emergency situations.

[0003] Existing emergency lighting control methods often have problems such as slow response speed, low intelligence, and low energy efficiency. These systems usually find it difficult to dynamically adjust lighting strategies according to different types and degrees of dangerous situations, resulting in insufficient lighting or energy waste in some cases. In addition, due to the lack of in-depth analysis of environmental data and multi-dimensional hazard assessment, it is difficult for existing systems to accurately judge the level of danger and formulate the optimal emergency lighting plan. On the other hand, with the rapid development of the Internet of Things and artificial intelligence technologies, new possibilities have been provided for the intelligent upgrade of emergency lighting control systems. However, how to effectively integrate multi-source sensor data, build an efficient dangerous situation analysis model, and how to achieve flexible and reliable lighting control in complex building environments are still technical problems that need to be solved urgently. Summary of the invention

[0004] The present invention provides a lighting control method and system in emergency situations, which realizes efficient utilization and precise control of lighting resources and ensures the durability and stability of emergency lighting.

[0005] In a first aspect, the present invention provides a lighting control method in an emergency situation, the lighting control method in an emergency situation comprising:

[0006] configuring a lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit;

[0007] monitoring the data of the environmental sensor by the control unit;

[0008] When the data of the environmental sensor indicates the existence of a dangerous situation, the control unit performs a dangerous situation analysis to obtain a dangerous level;

[0009] According to the danger level, the control unit generates an emergency lighting plan, wherein the emergency lighting plan includes information of lighting fixtures to be activated;

[0010] Based on the emergency lighting scheme, the control unit sends a control instruction to the lighting fixture to be activated;

[0011] In response to the control instruction, the lighting fixture to be activated switches to an emergency lighting state.

[0012] In a second aspect, the present invention provides a lighting control system in an emergency, the lighting control system in an emergency comprising:

[0013] A configuration module, used to configure a lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit;

[0014] A monitoring module, used for monitoring the data of the environmental sensor through the control unit;

[0015] An analysis module, configured to cause the control unit to analyze the dangerous situation and obtain a dangerous level when the data of the environmental sensor indicates that a dangerous situation exists;

[0016] A generating module, configured to generate an emergency lighting scheme by the control unit according to the hazard level, wherein the emergency lighting scheme includes information of lighting fixtures to be activated;

[0017] A sending module, configured to send a control instruction from the control unit to the lighting fixture to be activated based on the emergency lighting solution;

[0018] A switching module is used to switch the lighting fixture to be activated to an emergency lighting state in response to the control instruction.

[0019] In the technical solution provided by the present invention, the environmental state spatiotemporal graph is processed by a multi-branch fusion spatiotemporal graph convolutional network, which realizes the deep fusion analysis of multi-dimensional data such as smoke, temperature, carbon monoxide and electricity, and improves the accuracy and comprehensiveness of dangerous situation judgment. The optimal data transmission path is dynamically selected, combined with the dual communication interfaces of wired and wireless, to enhance the communication reliability and flexibility of the system in a complex network environment. The emergency lighting scheme is automatically generated and optimized based on the danger level, and the spatial distribution and power status of the lighting fixtures are considered to achieve efficient utilization and precise control of lighting resources. By setting and encrypting the control parameters of each lighting fixture to be activated separately, the security and pertinence of the system are improved, and it can adapt to the lighting needs of different areas. Intelligent switching and adaptive adjustment of lighting fixtures are realized, including power supply system switching, dynamic adjustment of brightness and power, and corresponding adjustment of the heat dissipation system, to ensure the durability and stability of emergency lighting. Through the state feedback mechanism, the control unit can grasp the working status of each lighting fixture in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] 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 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.

[0021] Figure 1 A schematic diagram of the steps of a lighting control method in an emergency situation according to an embodiment of the present invention;

[0022] Figure 2 Schematic diagram of the structure of a lighting control system in an emergency situation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] Embodiments of the present invention provide a lighting control method and system for emergency situations. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the lighting control method in an emergency situation in an embodiment of the present invention includes:

[0025] Step S1, configuring a lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit;

[0026] It is understandable that the execution subject of the present invention may be a lighting control system in an emergency, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0027] Specifically, the lighting control system is initialized first. The system includes multiple lighting fixtures, at least one environmental sensor, a central control unit, and a centralized power supply system. Each lighting fixture is equipped with a unique identification address, a 48V centralized power supply interface, a single-point power supply system with a built-in small battery, and an adjustable LED light source. The environmental sensor is used to detect fire, smoke or other dangerous situations. The central control unit has dual communication interfaces, wired and wireless (such as WiFi, Bluetooth), for data exchange with lighting fixtures and environmental sensors. The system administrator sets the power in normal working mode (30-40 watts) and the power in emergency mode (5-6 watts) for each lighting fixture through the user interface of the central control unit. At the same time, the administrator also needs to specify specific lamps that need to be kept lit in an emergency, and set the brightness level in emergency mode (such as 20% or 30%) for these lamps. The system performs self-tests to ensure that all components are working properly, including the centralized power supply system, single-point power supply system, communication interface, and environmental sensors.

[0028] Furthermore, multiple lighting fixtures are assigned addresses to ensure that each lighting fixture has a unique identification address, so that each lighting fixture can be independently controlled and identified. The lighting fixture power supply system is configured so that each lighting fixture has a 48V centralized power supply interface and a built-in battery to form a system with a single-point power supply capability. The dual power supply design helps to maintain the stability of the lighting system in an emergency. Even in the case of a centralized power supply interruption, the built-in battery can ensure continuous lighting. Each lighting fixture is configured with a light source, and each fixture is equipped with an adjustable LED light source to provide flexible brightness adjustment capabilities, so as to control the lighting effect according to environmental requirements. At the same time, the communication interface of the control unit is configured so that it can support both wired and wireless dual communication modes. The user can set the power of each lighting fixture's working mode through the user interface of the control unit, set the power value of the normal working mode and the power value of the emergency mode for each fixture, and ensure that the fixture can automatically adjust its power consumption as needed in different situations. Through the user interface of the control unit, specify which lamps enter the emergency mode, generate a list of emergency lighting fixtures, and set in advance which lamps need to be activated first in an emergency. Based on this list, the emergency mode brightness setting is performed for each emergency lighting fixture, and different emergency mode brightness values ​​are assigned to different fixtures to meet the lighting needs of different areas and environments. The entire lighting control system is self-checked to verify the functional status of all components, ensuring that each lighting fixture, environmental sensor and control unit are in normal working condition, ensuring the reliability and functionality of the system, and finally forming a fully configured lighting control system that can quickly respond and execute preset lighting plans in emergency situations, improving safety and system response efficiency.

[0029] Step S2, monitoring the data of the environmental sensor through the control unit;

[0030] Specifically, data is collected from environmental sensors to obtain raw environmental data from the external environment. The raw environmental data is classified to obtain smoke concentration data, temperature change rate data, carbon monoxide concentration data, and power supply status data, where each data type is processed accordingly according to its unique characteristics. The smoke concentration data is filtered to eliminate noise or abnormal values ​​in the collection process to obtain filtered smoke concentration data. At the same time, the temperature change rate data is calculated, and the real-time temperature change value is obtained by analyzing the temperature change in unit time to help identify sudden temperature increases. The carbon monoxide concentration data is standardized and compared with the predetermined standard value to obtain standardized carbon monoxide concentration data. The power supply status data is parsed to obtain key parameters of the power system, including voltage and current values. The analysis of power data helps to monitor the current power supply situation and determine whether the system is in a normal power supply state or whether there is a risk of power supply interruption. At the same time, the network status of the lighting control system is detected based on the preset communication protocol to determine whether the current communication method is wired communication or wireless communication. According to the determined current communication mode, the data transmission path is dynamically selected to ensure that the data can be transmitted through the optimal transmission path, reduce delays and improve the stability and efficiency of data transmission. The selected optimal data transmission path is suitable for the current communication mode and can be flexibly adjusted according to changes in network status to ensure efficient data transmission in various situations. Based on the optimal data transmission path, the filtered smoke concentration data, temperature change value per unit time, standardized carbon monoxide concentration data, voltage value and current value are periodically transmitted. Through the periodic transmission process, the data of the environmental sensor is monitored in real time, so that accurate environmental information can be provided in time when an emergency occurs, assisting the system to make emergency decisions and initiate corresponding lighting control measures.

[0031] Furthermore, the wired network and wireless network in the lighting control system are detected in parallel, and the state parameters of the wired and wireless networks are collected in real time, and the connection state, bandwidth, delay and other parameters related to the wired network, as well as the signal strength and connection stability of the wireless network are obtained respectively. Based on the state parameters of the wired network, the connection quality score of the wired network is calculated to obtain the score of the wired network, reflecting the overall performance and connection quality of the current wired network. At the same time, according to the state parameters of the wireless network, the signal strength and connection stability of the wireless network are calculated to generate the score of the wireless network. Through the comprehensive evaluation of the state of the wired and wireless networks, the connection quality of the two is compared and analyzed to determine the current optimal communication mode. Based on the current communication mode, the network topology diagram is analyzed, and the network structure model is drawn to display the nodes of the entire network system and their connection relationships. The network structure model is path planned to generate multiple candidate data transmission paths. Each path represents a possible data transmission route, and these paths are evaluated one by one through a preset path evaluation algorithm. According to the transmission efficiency of each candidate path, the transmission efficiency index of each path is calculated. The transmission efficiency index is comprehensively evaluated based on multiple factors, such as delay, bandwidth, data packet loss rate, etc. The evaluation process helps the system identify paths that have higher transmission efficiency under the current network conditions. Each candidate path is sorted according to its transmission efficiency index to generate a sorted path list. The paths in the list are arranged from high to low transmission efficiency. After the sorting is completed, the path with the highest transmission efficiency index is selected from the sorted path list as the optimal data transmission path. The optimal path is used for the current data transmission task to ensure that the lighting control system can exchange data with the highest transmission efficiency, improving the system's response speed and stability.

[0032] Step S3: When the data from the environmental sensor indicates that a dangerous situation exists, the control unit performs a dangerous situation analysis to obtain a dangerous level;

[0033] Specifically, the filtered smoke concentration data, the temperature change value per unit time, the standardized carbon monoxide concentration data, the voltage value and the current value in the environmental sensor data are reconstructed in time series to generate multi-dimensional time series data. Based on the multi-dimensional time series data, a spatiotemporal graph reflecting the environmental state is constructed. The spatiotemporal graph is a structured data that comprehensively represents environmental changes and can reflect the correlation between multiple sensor data in different time and space dimensions. The environmental state spatiotemporal graph is processed by a multi-branch fusion spatiotemporal graph convolution network. The network consists of multiple branches, including a smoke branch, a temperature branch, a carbon monoxide branch and a power branch, each of which specializes in processing corresponding sensor data. These branch networks are deployed in the control unit. In the smoke branch, the spatiotemporal graph is subjected to a time convolution operation to extract the smoke time features, and then the smoke space features are obtained through a graph convolution operation. The time features and spatial features are fused through a feature fusion layer to obtain the final smoke hazard features. The branch network consists of three layers of temporal convolutional layers and two layers of graph convolutional layers. The temporal convolutional layer uses 64 3x1 convolutional kernels and is activated by ReLU, while the graph convolutional layer uses 32 Chebyshev polynomial kernels and is activated by ELU. Finally, feature fusion is completed through gated linear units. At the same time, in the temperature branch, the temperature spatial features are extracted through graph convolution operations, and the temperature time features are obtained through temporal convolution operations. The two are fused to obtain temperature hazard features. The temperature branch network uses two layers of graph convolutional layers and three layers of temporal convolutional layers. The graph convolutional layer uses 48 Diffusion convolutional kernels, and the temporal convolutional layer uses 56 2x1 convolutional kernels, and uses LeakyReLU and GELU activation functions. The fusion layer realizes feature fusion through the self-attention mechanism. In the carbon monoxide branch, the temporal and spatial features of carbon monoxide are extracted respectively through parallel temporal convolution and graph convolution operations, and the two features are fused using the multi-head attention mechanism. This branch includes two parallel sub-branches. The temporal convolution sub-branch consists of three layers of 1D convolutional layers, each layer uses 40 convolution kernels and is activated by Swish. The graph convolution sub-branch uses two layers of spectral convolutional layers, uses 36 convolution kernels and uses Mish activation, and finally completes feature fusion through a 4-head multi-head attention mechanism. The power branch processes the spatiotemporal graph of the environmental state through a graph attention network to extract the spatial features of power. The gated recurrent unit (GRU) is used to process the temporal features of power. Finally, the spatial and temporal features are fused through an adaptive fusion layer to generate power hazard features. This branch consists of three layers of graph attention layers, each layer contains 6 attention heads, and uses residual connections and layer normalization. The two layers of GRU each contain 128 hidden units and are activated by Tanh. The adaptive fusion layer uses a dynamic weight generation network to fuse the features. The hazard features of smoke, temperature, carbon monoxide, and power are input into the multi-layer perceptron to extract the fused hazard features.The multi-layer perceptron consists of three fully connected layers, with 256, 128 and 64 hidden layer units respectively. Each layer is followed by batch normalization and Dropout layers, and the activation function uses ReLU. The fused danger features are classified by softmax, and the probability distribution of the danger level is output. If the probability distribution shows that there is a danger, a weighted voting analysis is performed based on the distribution to finally determine the current danger level.

[0034] Furthermore, the time dimension of the environmental state spatiotemporal graph is sliced ​​to obtain the time series data of carbon monoxide concentration. The time series data is input into the first 1D convolution layer, and 40 convolution kernels are used for convolution operation to generate the first layer of time feature graph. In order to improve the nonlinear expression ability of the model, the Swish activation function is applied to the first layer of time feature graph to obtain the activated time feature graph to ensure the continuity and fluency of feature extraction. The activated time feature graph of the first layer is input into the second and third 1D convolution layers in turn, each layer uses 40 convolution kernels, and is processed by the Swish activation function. After multiple layers of convolution and activation operations, the time characteristics of carbon monoxide are extracted to fully capture the change law of data in the time dimension. At the same time, the space dimension of the environmental state spatiotemporal graph is sliced ​​to obtain the spatial distribution data of carbon monoxide concentration. The spatial distribution data is input into the first layer of graph convolution layer, and 36 spectral convolution kernels are used for graph convolution operation to generate the first layer of spatial feature graph. In order to enhance the expression ability of the feature, the Mish activation function is applied to process the first layer of spatial feature graph to obtain the activated spatial feature graph. The activated spatial feature map is input into the second graph convolution layer, convolved by 36 spectral convolution kernels, and the Mish activation function is applied to finally generate the spatial features of carbon monoxide. Through the graph convolution operation, the distribution characteristics of carbon monoxide in the spatial dimension are accurately captured. The temporal features of carbon monoxide and the spatial features of carbon monoxide are concatenated to obtain the spatiotemporal joint features. The multi-head attention mechanism is applied to the spatiotemporal joint features, and the self-attention weights are calculated using 4 attention heads. Through the weighted fusion of the attention mechanism, the features that are more important for the prediction of dangerous situations are focused on, and the dangerous features of carbon monoxide are obtained.

[0035] Step S4: The control unit generates an emergency lighting plan according to the danger level, wherein the emergency lighting plan includes information of the lighting fixtures to be activated;

[0036] Specifically, a threshold judgment is performed on the detected danger level, and different danger levels are distinguished according to the set threshold, and a danger level classification result is obtained. Different danger levels indicate the risk level in the current environment. Based on the danger level classification result, a basic lighting strategy corresponding to the current danger level is selected from the preset emergency lighting strategy library to generate a first lighting plan. A spatial distribution analysis is performed on each lighting fixture in the lighting control system, and the geographical distribution and lighting range of each lighting fixture are understood by analyzing the location information of the lighting fixture and the area it covers. Based on the location information and the coverage area information, the first lighting plan is optimized, and the activation sequence and area of ​​the lighting fixtures are adjusted so that each area can obtain the best lighting effect when a danger occurs, and a more reasonable second lighting plan is generated. The power of each lighting fixture to be activated in the second lighting plan is evaluated to ensure that the lighting fixture can continue to work in an emergency state, especially in the case of possible power outages or insufficient power. By evaluating the power status of each lamp, it is identified which lamps have sufficient power and which lamps may need to be adjusted or replaced first. Based on the power status information, the second lighting plan is adjusted accordingly to ensure that the emergency lighting plan can play the maximum effect in actual situations. Specifically, lamps with sufficient power will be activated first, while lamps with insufficient power may be replaced with other backup lamps or have their usage time reduced to ensure the stable operation of the entire system. Through adjustments, the final emergency lighting plan includes which lamps need to be activated, and ensures that they have sufficient power and functions to cope with dangerous situations, ensuring lighting coverage and continuity in emergency situations.

[0037] Step S5: Based on the emergency lighting solution, the control unit sends a control instruction to the lighting fixture to be activated;

[0038] Specifically, the emergency lighting scheme is parsed to obtain a list of lighting fixtures to be activated, and the control parameters of each lighting fixture to be activated are determined. The control parameters generally include specific information such as the brightness of the lighting fixture, the lighting mode, and the activation duration. Based on the list of lighting fixtures to be activated, the current communication status of the lighting control system is detected to ensure smooth communication, and a list of candidate communication modes is generated, including selectable communication modes, such as wired communication and wireless communication. According to the current system conditions, it is determined which mode to use for data transmission. According to the list of candidate communication modes, the lighting fixtures to be activated are grouped into wired communication groups and wireless communication groups. The lamps in the wired communication group are controlled through physical lines such as network cables, while the lamps in the wireless communication group are controlled through wireless channels. The lamps in the wired communication group are address-encoded so that each lamp can identify and receive instructions through a unique address to form a wired communication instruction sequence. Similarly, for the lamps in the wireless communication group, channel allocation is performed to ensure the stability and non-interference of the wireless channel, and a wireless communication instruction sequence is generated. The wired communication instruction sequence and the wireless communication instruction sequence are merged to obtain a target control instruction set. In order to improve the security of data transmission, the target control instruction set is encrypted to prevent external attacks or data tampering during instruction transmission, ensuring the reliability of emergency lighting instructions. Based on the preset communication protocol, the encrypted control instructions are sent to the corresponding lighting fixtures to be activated. The instructions of the wired communication group are sent through the network cable, while the instructions of the wireless communication group will be transmitted through the allocated wireless channel. After the instructions arrive, the lighting fixtures switch to the emergency lighting state according to the received instructions, ensuring that the lighting needs in emergency situations are responded to and executed quickly and accurately.

[0039] Step S6: In response to the control instruction, the lighting fixture to be activated switches to the emergency lighting state.

[0040] Specifically, the received control instructions are decrypted to obtain the decrypted control parameters, which include the operation settings of the lighting fixtures in the emergency state, such as power supply mode switching, brightness adjustment and power control. Based on the decrypted control parameters, the power supply system of the lighting fixture is switched, and the lighting fixture is switched from the 48V centralized power supply interface that it originally relied on to the single-point power supply system with built-in batteries, ensuring that even if the centralized power supply is interrupted, the lighting fixture can still keep working through the built-in battery to meet the needs of emergency lighting. According to the brightness settings in the decrypted control parameters, the driving current of the LED light source is adjusted to ensure that the brightness output by the lighting fixture reaches the set target value. At the same time, according to the power setting in the control parameters, the power control circuit of the lighting fixture is adjusted accordingly, and the lighting fixture is switched from the power value in the normal working mode to the power value in the emergency mode, effectively reducing energy consumption and ensuring that the duration of emergency lighting can be extended under limited power conditions. In order to ensure the stability of emergency lighting, the built-in battery of the lighting fixture is tested for power and the current power status is obtained. Based on the current power status and the pre-set emergency lighting duration requirements, the brightness of the LED light source is dynamically adjusted to achieve the best balance between power and lighting duration, ensuring that the lighting fixtures can continue to work for a long enough time in an emergency. Through dynamic optimization and adjustment of brightness, the stability of lighting is ensured and premature extinguishing due to insufficient power is avoided. At the same time, based on the optimized brightness output, the heat dissipation system of the lighting fixture is adjusted. Since the lamps may work for a long time in an emergency, the adjustment of the heat dissipation system helps to ensure that the performance or life of the lamps will not be affected by excessive temperature during continuous operation. By optimizing the heat dissipation system, the lamps can maintain a stable lighting working state in an emergency. The emergency lighting working state of the lighting fixture is monitored in real time, and the status information is fed back to the control unit through a preset communication protocol. Through the feedback mechanism, the control unit can understand the working status of each lighting fixture in real time to ensure the normal operation and troubleshooting of the system during the entire emergency lighting period.

[0041] In the embodiment of the present invention, the environment state spatiotemporal graph is processed by a multi-branch fusion spatiotemporal graph convolutional network, which realizes the deep fusion analysis of multi-dimensional data such as smoke, temperature, carbon monoxide and electricity, and improves the accuracy and comprehensiveness of dangerous situation judgment. The optimal data transmission path is dynamically selected, combined with the wired and wireless dual communication interfaces, to enhance the communication reliability and flexibility of the system in a complex network environment. The emergency lighting scheme is automatically generated and optimized based on the danger level, and the spatial distribution and power status of the lighting fixtures are considered to achieve efficient utilization and precise control of lighting resources. By setting and encrypting the control parameters of each lighting fixture to be activated separately, the security and pertinence of the system are improved, and it can adapt to the lighting needs of different areas. Intelligent switching and adaptive adjustment of lighting fixtures are realized, including power supply system switching, dynamic adjustment of brightness and power, and corresponding adjustment of the heat dissipation system, to ensure the durability and stability of emergency lighting. Through the state feedback mechanism, the control unit can grasp the working status of each lighting fixture in real time.

[0042] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0043] Assign addresses to multiple lighting fixtures to obtain a unique identification address for each lighting fixture, and configure the power supply system for multiple lighting fixtures so that each lighting fixture has a 48V centralized power supply interface and a single-point power supply system with a built-in battery;

[0044] The light sources of the plurality of lighting fixtures are configured so that each lighting fixture has an adjustable LED light source, and the communication interface of the control unit is configured so that the control unit has a dual communication interface of wired and wireless;

[0045] Based on the user interface of the control unit, the working mode power of each lighting fixture is set to obtain a normal working mode power value and an emergency mode power value, and based on the user interface of the control unit, the emergency mode of the lighting fixture is specified to obtain an emergency lighting fixture list;

[0046] Based on the user interface of the control unit, the emergency mode brightness of the lamps in the emergency lighting fixture list is set to obtain the emergency mode brightness value of each emergency lighting fixture, and the lighting control system is self-checked to verify the functional status of all components to obtain a configured lighting control system.

[0047] Specifically, addresses are assigned to multiple lighting fixtures to ensure that each fixture can be independently identified and controlled, and a unique identification address is generated for each lighting fixture. The address is generated using a preset logical rule based on the physical location or installation order of the lighting fixture through an allocation algorithm in the control unit. For example, if there are N lighting fixtures, the address of each lighting fixture can be expressed as ,in is the index of the lamp, and the value range is 1 to N. Configure the power supply system for each lamp. Each lamp needs to have both a 48V centralized power supply interface and a single-point power supply system with a built-in battery. The dual power supply design ensures that under normal circumstances, the lamp works through the centralized power supply system, reducing dependence on individual batteries, but in an emergency, when the centralized power supply system fails, the lamp can automatically switch to the built-in battery power supply to continue to provide power for emergency lighting. The switching process is usually automatically managed by the control unit, and dynamic switching is performed by detecting the voltage status of the main power supply. When the main power supply voltage Below a preset threshold , that is, when When the system is powered by the built-in battery, the system will automatically switch to power supply by the built-in battery. After completing the configuration of the power supply system, configure the light source of each lamp to ensure that each lamp is equipped with an adjustable LED light source, and the brightness of the LED light source can be adjusted according to demand, so that the lighting system can flexibly respond to lighting needs in different application scenarios. By adjusting the drive current , control the brightness of the LED light source , which is related by the formula Indicates that Is a constant, which is related to the luminous efficiency of the LED. The control system only needs to adjust the driving current to achieve the required brightness output. Configure the communication interface of the control unit. The control unit must have both wired and wireless communication interfaces to ensure that the system can perform efficient data transmission in different communication environments. For wired communication, it is carried out through Ethernet or other physical lines, while wireless communication is implemented through protocols such as Wi-Fi or ZigBee. Through these two interfaces, it is ensured that when a communication failure occurs or a certain communication method is limited, data can still be transmitted through the backup communication method, thereby ensuring the stability and reliability of the lighting control system. Set the working mode power of each lighting fixture through the user interface of the control unit. Through this user interface, the operator sets the normal working mode power value and emergency mode power value of each lamp. These power values ​​determine the energy consumption status of the lamp during normal lighting and emergency lighting. Normal working mode power value Usually higher to meet daily lighting needs, while the emergency mode power value The power setting formula is expressed as ,in is the supply voltage, is the working current of the lamp. , control power from Switch to , to adapt to different lighting needs. After setting the working mode power, specify the emergency mode for each lighting fixture through the user interface of the control unit to generate a list of emergency lighting fixtures. The list includes lighting fixtures that need to be activated first in an emergency. The selection of emergency lighting fixtures is based on factors such as the layout of the building and the evacuation routes of personnel to ensure that key areas always have sufficient lighting. At this time, according to the needs of emergency lighting, mark which lamps enter emergency mode, and ensure that the power and brightness settings of these lamps meet the requirements of the emergency state. Set the brightness of the emergency mode lamps based on the user interface of the control unit. The brightness value of each emergency lamp Dynamically adjust according to the severity of the emergency and the battery power. The brightness value is achieved by adjusting the LED drive current to ensure that sufficient lighting is provided in an emergency and the built-in battery life is extended. For example, in some key areas, such as escape routes, the brightness is set to a higher value, while in secondary areas the brightness is appropriately reduced to save power. After all settings are completed, the lighting control system is self-checked to verify the functional status of each component. The self-check process includes a comprehensive check of the lighting fixture's power supply system, LED light source, communication interface, control unit, etc. to ensure that each component is operating normally. By sending test instructions, simulate the operation in an emergency situation and confirm whether the lamp can successfully switch to emergency mode. And maintain the preset power and brightness output.

[0048] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0049] Collect data from environmental sensors to obtain raw environmental data, and classify the raw environmental data to obtain smoke concentration data, temperature change rate data, carbon monoxide concentration data, and power supply status data;

[0050] The smoke concentration data is filtered to obtain the filtered smoke concentration data; the temperature change rate data is calculated to obtain the temperature change value per unit time; the carbon monoxide concentration data is standardized to obtain the standardized carbon monoxide concentration data; the power supply status data is analyzed to obtain the voltage value and the current value;

[0051] Based on the preset communication protocol, the network status of the lighting control system is detected to obtain the current communication mode, and the data transmission path is dynamically selected according to the current communication mode to obtain the optimal data transmission path;

[0052] Based on the optimal data transmission path, the filtered smoke concentration data, the temperature change value per unit time, the standardized carbon monoxide concentration data, the voltage value and the current value are periodically transmitted to obtain the data of the environmental sensor.

[0053] Specifically, various raw environmental data monitored by environmental sensors are collected. These data come from multiple different sensor modules, such as sensors for monitoring smoke concentration, temperature sensors, carbon monoxide concentration sensors, and power sensors for detecting power supply status. By collecting raw environmental data, the overall state of the current environment is obtained. The raw data is classified into smoke concentration data, temperature change rate data, carbon monoxide concentration data, and power supply status data. The collected smoke concentration data is filtered to remove noise and outliers in the collection process. Filtering is achieved by applying a low-pass filter or a moving average filter. Assume that the time series of the smoke concentration data is After being processed by the filter, the filtered smoke concentration data is obtained For example, the moving average filter is calculated by the following formula: ;

[0054] in, is the size of the filter window, Indicates the smoke concentration data at the previous moment. In this way, the sensor data is smoothed, short-term fluctuations are removed, and a more reliable concentration value is obtained. The temperature change rate data is calculated to measure the temperature change amplitude per unit time. Assume that the function of temperature change over time is , the temperature change rate is expressed as the time derivative. In the case of discrete data, the temperature change rate is calculated by the following formula: ;

[0055] in, is the temperature change rate, and Represent the temperature at the current time and the previous time respectively, is the time interval. Through calculation, the rapid change of temperature in the environment is detected to determine whether there is an abnormal temperature rise, which may indicate a fire or overheating of the equipment. The carbon monoxide concentration data is standardized to convert the original concentration data into a more comparable data range for subsequent analysis. Standardization is achieved through the following formula: ;

[0056] in, represents the normalized carbon monoxide concentration, is the original concentration, is the average carbon monoxide concentration, is the standard deviation. Through standardization, the values ​​of different intervals are unified to the same scale, which is convenient for subsequent analysis and comparison in different scenarios. At the same time, the power supply status data is analyzed to extract the voltage and current values ​​in the power system. The power status is read from the sensor data and decomposed into voltage and current values ​​through the analysis algorithm. and current Two key parameters. These parameters are important indicators to measure the power supply status, such as: ;

[0057] in, is power, is the voltage, The current is measured by the current source. Through analysis, the health status of the power supply is monitored in real time to determine whether there is abnormal fluctuation or interruption in the power supply. The current network status is detected to ensure smooth data transmission. The optimal data transmission path is selected to ensure stable transmission of environmental data. Through the preset communication protocol, the availability, signal strength and stability of the current network are detected to obtain a list of candidate communication methods. Assuming that the current network status includes wired communication and wireless communication, the optimal transmission path is selected based on the detection results. A dynamic selection algorithm is used. For example, by evaluating the delay of each communication path and bandwidth , calculate the path priority. Path priority It is expressed as: ;

[0058] In this formula, The higher the value, the higher the transmission efficiency of the path, so it is preferred. In this way, the path with the smallest communication delay and the largest bandwidth is selected to achieve the best data transmission. Periodic data transmission is performed through the optimal data transmission path. Smoke concentration data after filtering is periodically transmitted , Temperature change rate per unit time , Standardized carbon monoxide concentration , and the voltage obtained by analysis and current Through periodic transmission, it monitors environmental changes in real time and responds promptly when an emergency occurs.

[0059] In a specific embodiment, the execution step detects the network status of the lighting control system based on a preset communication protocol, obtains the current communication mode, and dynamically selects the data transmission path according to the current communication mode. The process of obtaining the optimal data transmission path may specifically include the following steps:

[0060] Conduct parallel detection on the wired network and wireless network in the lighting control system to obtain wired network status parameters and wireless network status parameters;

[0061] Based on the wired network status parameters, the connection quality score of the wired network is calculated to obtain a wired network score, and based on the wireless network status parameters, the signal strength and stability of the wireless network is calculated to obtain a wireless network score;

[0062] Compare and analyze the wired network scores and the wireless network scores to obtain the current communication mode, and analyze the network topology diagram based on the current communication mode to obtain the network structure model;

[0063] Performing path planning on the network structure model to obtain multiple candidate data transmission paths, and evaluating the multiple candidate data transmission paths based on a preset path evaluation algorithm to obtain a transmission efficiency index of each candidate data transmission path;

[0064] According to the transmission efficiency index, multiple candidate data transmission paths are sorted to obtain a sorted path list, and the path with the highest transmission efficiency index is selected from the sorted path list as the optimal data transmission path.

[0065] Specifically, the wired network and wireless network in the lighting control system are detected in parallel, and the status parameters of the wired network and the wireless network are collected at the same time to ensure that the performance of the network under different communication modes can be evaluated. For the wired network, the network bandwidth, delay, jitter, packet loss rate and other parameters are monitored. For the wireless network, the signal strength, connection stability, interference strength and other information are evaluated. Based on the collected wired network status parameters, the connection quality score of the wired network is calculated. Assume that the network status parameters include bandwidth , delay D and packet loss rate , by comprehensively considering these factors to calculate the network connection quality score The network scoring formula is expressed as:

[0066] ;

[0067] in, Indicates the network bandwidth (Mbps), represents the network delay (ms), and The greater the bandwidth, the lower the latency and packet loss rate, and the higher the network score. The higher the score, the better the connection quality of the wired network. Similarly, for wireless networks, the signal strength and connection stability are calculated based on the wireless network status parameters. Wireless Network Score By signal strength and network stability Calculation. The signal strength of a wireless network is usually measured in dBm, while network stability is assessed by monitoring the persistence of the connection and signal fluctuations. The formula for wireless network scoring is expressed as: ;

[0068] in, is the signal strength, It is a coefficient based on the stability of the connection, ranging from 0 to 1. The stronger the signal strength, the higher the connection stability, and the network score The higher the wired network score is, the higher the wireless network score is. and , determine the current optimal communication method. > , wired network is used for data transmission first; otherwise, wireless network is selected. Based on the current communication method, a topology diagram is analyzed for the entire network to understand the structure of the network and the connection relationship between each node. Network topology diagram analysis can help the system understand how each device in the lighting control system is connected to each other, including the communication links and physical locations between devices. The generation of the topology diagram is achieved through the topology discovery algorithm, which automatically draws all nodes and edges (i.e. devices and connection lines) in the network. Path planning is performed on the generated network structure model. Multiple candidate data transmission paths are found in the topology diagram, and each path is a potential communication route. Through path planning, multiple candidate paths are generated, which represent different data transmission methods from the control unit to the lighting fixture. In order to evaluate the transmission efficiency of each candidate path, these paths are evaluated based on the preset path evaluation algorithm. Assume that the transmission efficiency of each path is related to the delay. ,bandwidth and hop count Related, transmission efficiency index Calculated by the following formula: ;

[0069] in, represents the bandwidth of the path, represents the total delay of the path, is the number of hops in the path (i.e. the number of intermediate nodes). The larger the bandwidth, the lower the delay, and the fewer the hops, the higher the transmission efficiency index of the path. The higher the transmission efficiency, the higher the transmission efficiency. According to the transmission efficiency index, multiple candidate data transmission paths are sorted. The path with the highest transmission efficiency index is selected from the sorted path list as the optimal data transmission path.

[0070] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0071] Reconstruct the filtered smoke concentration data, the temperature change value per unit time, the standardized carbon monoxide concentration data, the voltage value and the current value in the environmental sensor data to obtain multi-dimensional time series data;

[0072] Construct a spatiotemporal graph based on multidimensional time series data to obtain a spatiotemporal graph of environmental states;

[0073] The environment state spatiotemporal graph is processed by a multi-branch fusion spatiotemporal graph convolutional network to obtain a dangerous situation feature vector, wherein the multi-branch fusion spatiotemporal graph convolutional network includes a smoke branch, a temperature branch, a carbon monoxide branch, and a power branch; the multi-branch fusion spatiotemporal graph convolutional network is deployed in the control unit;

[0074] In the smoke branch, a temporal convolution operation is performed on the environment state spatiotemporal graph to obtain the smoke temporal features, and a graph convolution operation is performed on the smoke temporal features to obtain the smoke spatial features. Then, the smoke temporal features and the smoke spatial features are fused to obtain the smoke hazard features. The smoke branch includes: three layers of temporal convolution layers, each layer uses 64 3x1 convolution kernels and adopts the ReLU activation function; two layers of graph convolution layers, each layer uses 32 Chebyshev polynomial kernels and adopts the ELU activation function; a feature fusion layer, which uses a gated linear unit for feature fusion;

[0075] In the temperature branch, the environment state spatiotemporal graph is convolved to obtain the temperature spatial feature, and the temperature spatial feature is convolved to obtain the temperature time feature. Then, the temperature spatial feature and the temperature time feature are fused to obtain the temperature danger feature. The temperature branch includes: two graph convolution layers, each layer uses 48 Diffusion convolution kernels and adopts LeakyReLU activation function; three temporal convolution layers, each layer uses 56 2x1 convolution kernels and adopts GELU activation function; one feature fusion layer, which adopts self-attention mechanism for feature fusion;

[0076] In the carbon monoxide branch, the environmental state spatiotemporal graph is subjected to parallel temporal convolution and graph convolution operations to obtain the carbon monoxide temporal features and carbon monoxide spatial features respectively, and then the carbon monoxide temporal features and carbon monoxide spatial features are fused by the attention mechanism to obtain the carbon monoxide hazard features; the carbon monoxide branch includes: two parallel sub-branches, the temporal convolution sub-branch contains three layers of 1D convolution layers, each layer uses 40 convolution kernels and adopts the Swish activation function; the graph convolution sub-branch contains two layers of graph convolution layers, each layer uses 36 spectral convolution kernels and adopts the Mish activation function; a multi-head attention layer for feature fusion, which contains 4 attention heads;

[0077] In the power branch, the environment state spatiotemporal graph is processed by graph attention network to obtain power spatial features, and the power spatial features are processed by gated recurrent units to obtain power temporal features, and then the power spatial features and power temporal features are adaptively fused to obtain power hazard features; the power branch includes: three layers of graph attention layers, each layer contains 6 attention heads, using residual connections and layer normalization; two layers of gated recurrent units, each layer contains 128 hidden units, using Tanh activation function; an adaptive fusion layer, using a dynamic weight generation network for feature fusion;

[0078] The smoke hazard features, temperature hazard features, carbon monoxide hazard features and electricity hazard features are input into a multi-layer perceptron to obtain a fusion hazard feature; the multi-layer perceptron includes three fully connected layers, the number of hidden layer units is 256, 128 and 64 respectively, each layer is followed by a batch normalization and a Dropout layer, and the activation function uses ReLU;

[0079] The fused danger features are classified by softmax to obtain the danger level probability distribution. When the danger level probability distribution indicates the existence of a dangerous situation, a weighted voting analysis is performed based on the danger level probability distribution to obtain the danger level.

[0080] Specifically, the filtered smoke concentration data, temperature change value per unit time, standardized carbon monoxide concentration data, voltage value and current value in the environmental sensor data are reconstructed to obtain multidimensional time series data. These data are rearranged in chronological order to form a multidimensional time series data matrix. Assume that the acquisition time is , smoke concentration, temperature change, carbon monoxide concentration, voltage and current are recorded as , the multidimensional time series matrix is ​​expressed as:

[0081] ;

[0082] in, is a The time series matrix represents the multidimensional environmental data at different time points. The dynamic changes of each data in time are captured through the multidimensional time series data matrix. A spatiotemporal graph is constructed based on the multidimensional time series data to generate a spatiotemporal graph of the environmental state. The construction of the spatiotemporal graph is usually based on the mutual relationship between time nodes and the spatial association between different sensor data. The nodes of the spatiotemporal graph represent the state of the environmental data, while the edges represent the temporal or spatial association between different nodes. The spatiotemporal graph of the environmental state is processed by a multi-branch fusion spatiotemporal graph convolutional network. The convolutional network is mainly divided into a smoke branch, a temperature branch, a carbon monoxide branch, and a power branch. Each branch is responsible for processing the corresponding environmental data and finally fused into a comprehensive dangerous situation feature vector. The multi-branch fusion spatiotemporal graph convolutional network deployed in the control unit extracts spatiotemporal features by performing time convolution and graph convolution operations on different data of each branch. In the smoke branch, the spatiotemporal graph of the environmental state is subjected to a time convolution operation to extract the dynamic features of the smoke concentration data at different time points. Through three layers of temporal convolution layers, each layer uses 64 3x1 convolution kernels to fully process the smoke concentration data in the temporal dimension. The convolution kernel size of 3x1 means that three adjacent data points are taken in time for convolution. After each layer of temporal convolution, the ReLU activation function is applied to preserve the nonlinear features. After the temporal convolution processing, the extracted smoke temporal features are subjected to graph convolution operations to extract the spatial features of the data. Through two layers of graph convolution layers, each layer uses 32 Chebyshev polynomial kernels to capture the spatial associations between different sensor data. The Chebyshev polynomial kernel is suitable for graph convolution operations and can effectively process graph structure data in non-Euclidean space. After the convolution operation, the ELU activation function is applied to enhance the nonlinear expression ability of the data. The temporal and spatial features of the smoke branch are fused through the feature fusion layer. The fusion layer uses the gated linear unit (GRU) for feature fusion to obtain the final smoke hazard features. In the temperature branch, the environmental state spatiotemporal graph is subjected to graph convolution operations to extract the spatial features of temperature changes. Through two layers of graph convolution layers, each layer uses 48 diffusion convolution kernels to capture the spatial correlation of temperature change rate. Diffusion convolution is a convolution method suitable for processing time series data, which can simulate the diffusion process of data on the graph. The temporal characteristics of temperature change are extracted through three layers of temporal convolution layers, each layer uses 56 2x1 convolution kernels and uses GELU activation function for processing. Through this structure, the trend of temperature change in the time dimension is effectively analyzed. The spatial and temporal features of temperature are fused through the self-attention mechanism to obtain the temperature hazard features. In the carbon monoxide branch, the environmental state spatiotemporal graph is subjected to parallel temporal convolution and graph convolution operations to extract the temporal and spatial features of carbon monoxide respectively. The temporal convolution sub-branch contains three layers of 1D convolution layers, each layer uses 40 convolution kernels and uses Swish activation function.Swish is a smooth activation function that can improve the performance of the network. The graph convolution sub-branch contains two layers of graph convolution layers, each layer uses 36 spectral convolution kernels and adopts Mish activation function. Spectral convolution kernels can capture the frequency domain characteristics of graph structure data, and Mish activation function has good gradient propagation performance. The carbon monoxide hazard feature is obtained by fusing the temporal and spatial features of carbon monoxide through the multi-head attention mechanism. The multi-head attention mechanism contains 4 attention heads, which can focus on different feature dimensions in parallel and enhance the effect of feature fusion. In the power branch, the environmental state spatiotemporal graph is processed by the graph attention network to extract the spatial features of power data. The graph attention network adaptively assigns weights to different features through a structure with 6 attention heads in each layer. The temporal features of power data are extracted using two layers of gated recurrent units, each layer contains 128 hidden units, and uses Tanh activation function, which can effectively process long dependency information in time series. The power branch fuses the temporal features and spatial features through the adaptive fusion layer to obtain the power hazard feature. The adaptive fusion layer uses a dynamic weight generation network to perform weighted fusion according to the importance of the features. After completing the feature extraction of each branch, the smoke hazard features, temperature hazard features, carbon monoxide hazard features, and electricity hazard features are input into the multilayer perceptron for feature fusion. The multilayer perceptron contains three fully connected layers, with 256, 128, and 64 hidden layer units, respectively. Batch normalization and Dropout processing are performed after each fully connected layer to prevent the model from overfitting, and ReLU is used as the activation function to enhance the nonlinear representation ability of the model. The fused features are passed through the softmax classification layer to output a probability distribution of the hazard level. If the softmax classification result shows that there is a high probability of a hazard level, a weighted voting analysis is performed based on the probability distribution to finally determine the current hazard level. In this way, multi-dimensional environmental data is comprehensively analyzed to make an accurate assessment of the hazard situation and trigger corresponding emergency response measures when necessary.

[0083] In a specific embodiment, the execution step performs parallel time convolution and graph convolution operations on the environmental state spatiotemporal graph in the carbon monoxide branch to obtain carbon monoxide time features and carbon monoxide spatial features respectively, and then fuses the carbon monoxide time features and carbon monoxide spatial features by an attention mechanism to obtain the carbon monoxide hazard features. The process may specifically include the following steps:

[0084] Slice the environmental state space-time diagram in the time dimension to obtain the carbon monoxide concentration time series data;

[0085] The carbon monoxide concentration time series data is input into the first 1D convolution layer, and 40 convolution kernels are used for convolution operation to obtain the first layer time feature map, and the first layer time feature map is processed by Swish activation function to obtain the first layer time feature map after activation;

[0086] The time feature map after activation of the first layer is input into the second and third 1D convolution layers in sequence. Each layer uses 40 convolution kernels for convolution operation and applies the Swish activation function to obtain the carbon monoxide time feature. The environmental state spatiotemporal map is sliced ​​in the spatial dimension to obtain the spatial distribution data of carbon monoxide concentration.

[0087] The spatial distribution data of carbon monoxide concentration is input into the first graph convolution layer, and 36 spectral convolution kernels are used to perform graph convolution operation to obtain the first-layer spatial feature map, and the first-layer spatial feature map is processed by Mish activation function to obtain the spatial feature map after activation of the first layer;

[0088] The spatial feature map after activation in the first layer is input into the second layer of graph convolution layer, and 36 spectral convolution kernels are used for graph convolution operation and Mish activation function is applied to obtain the spatial feature of carbon monoxide, and the temporal feature of carbon monoxide and the spatial feature of carbon monoxide are concatenated to obtain the spatiotemporal joint feature;

[0089] Multi-head attention processing is performed on the spatiotemporal joint features, and four attention heads are used to calculate the self-attention weights to obtain the carbon monoxide hazard features.

[0090] Specifically, the time dimension slice operation is performed on the environmental state space-time graph. The environmental state space-time graph contains data on the changes in carbon monoxide concentration over time and space. The slicing operation is used to extract the sequence data of carbon monoxide concentration in the time dimension. Assume that at time point The carbon monoxide concentration is recorded as , through time slicing, we can get the time series data of carbon monoxide concentration: ;

[0091] The carbon monoxide concentration time series data is input into the first 1D convolution layer. This layer uses 40 convolution kernels to perform convolution operations to extract local features between consecutive time points in the time series. Assume that the convolution kernel size is , then the convolution operation is expressed as: ;

[0092] in, It's at the time The convolution result calculated above is It is The weight of the convolution kernel, It is before The carbon monoxide concentration value at a time point. Through the convolution operation, the changing trend of carbon monoxide concentration in the time series is captured. After the convolution operation, the Swish activation function is applied to process the first layer of convolution results. The Swish activation function can be calculated by the following formula: ;

[0093] in, is the Sigmoid function. The characteristic of the Swish activation function is that it can retain the characteristic information of negative inputs, making the model perform more smoothly and stably when processing complex environmental data. After Swish activation, the time feature map after the first layer is activated is obtained. The time feature map after the first layer is activated is input into the second and third 1D convolution layers in turn. Each layer still uses 40 convolution kernels and is processed by the Swish activation function. After multiple layers of convolution operations, the deep time features are extracted from the time series to obtain the time features of carbon monoxide, which describes the complex pattern of carbon monoxide concentration changing over time, such as its growth rate, change amplitude, etc. At the same time, the spatial dimension of the environmental state space-time diagram is sliced ​​to extract the spatial distribution data of carbon monoxide concentration. Assuming that at different locations The carbon monoxide concentrations are , through the spatial slicing operation, the spatial distribution data of carbon monoxide is obtained:

[0094] ;

[0095] The spatial distribution data is input into the first graph convolution layer. The graph convolution layer uses 36 spectral convolution kernels for convolution operation. Spectral convolution is a convolution method suitable for processing non-Euclidean space (such as graph structure data), which can effectively capture the correlation between different nodes (positions). Spectral convolution performs convolution operation in the frequency domain by eigendecomposing the Laplacian matrix of the graph. The convolution result is expressed as: ;

[0096] in, Is the spatial position The convolution result is is the weight parameter of the spectral convolution kernel. After this convolution operation, the mutual influence between different spatial positions and the spatial diffusion pattern of carbon monoxide concentration are obtained. After the convolution operation, the Mish activation function is applied to the first layer of spatial feature map. The Mish activation function has good gradient fluidity and stability, and its formula is: ;

[0097] After Mish activation processing, the spatial feature map after activation of the first layer is obtained. The spatial feature map after activation of the first layer is input into the second layer of graph convolution layer, and 36 spectral convolution kernels are used for convolution operation, and the Mish activation function is continued to be applied to obtain the spatial characteristics of carbon monoxide, which reflects the distribution pattern of carbon monoxide concentration in different locations and the spatial correlation between them. The temporal characteristics and spatial characteristics of carbon monoxide are spliced ​​to obtain the spatiotemporal joint characteristics, so that the system can simultaneously consider the interaction between time and space. The spatiotemporal joint features are processed by a multi-head attention mechanism. The multi-head attention mechanism calculates the self-attention weights through multiple attention heads in parallel, so that the system can focus on the most important information in different feature dimensions. Each attention head calculates the weight according to the input spatiotemporal characteristics and sums them weightedly. Assume that the input of the attention mechanism is the spatiotemporal joint feature , then the attention score Calculated by the following formula: ;

[0098] in, are the query, key and value of the spatiotemporal joint feature, respectively. is the dimension of the feature vector. After being processed by the multi-head attention mechanism, the weights of different features are adaptively assigned to obtain the dangerous features of carbon monoxide.

[0099] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0100] Perform threshold judgment on the danger level to obtain the danger level classification result;

[0101] Based on the hazard level classification result, a corresponding basic lighting strategy is selected from a preset emergency lighting strategy library to obtain a first lighting plan;

[0102] Perform spatial distribution analysis on lighting fixtures in the lighting control system to obtain location information and coverage area information of lighting fixtures;

[0103] Based on the location information and the coverage area information, the first lighting scheme is optimized to obtain a second lighting scheme;

[0104] Performing power evaluation on each lighting fixture to be activated in the second lighting scheme to obtain power status information;

[0105] The second lighting plan is adjusted according to the power status information to obtain an emergency lighting plan, wherein the emergency lighting plan includes information about the lighting fixtures to be activated.

[0106] Specifically, a threshold is determined for the current danger level. Environmental data is classified according to the preset danger level threshold. For example, assuming that the danger level data provided by the environmental monitoring system is based on the smoke concentration, temperature change, carbon monoxide concentration and other parameters collected by the sensor, the danger level is expressed as , whose value range is , where 0 means no danger and 1 means extremely high danger. and To determine the risk level. , it is a low-risk state; if , it is a medium-risk state; and when , it indicates a high-risk state. Through this process, the current risk level is quickly divided into three categories: low, medium, and high. Based on the risk level classification results, select the basic lighting strategy corresponding to the current risk level from the preset emergency lighting strategy library. The emergency lighting strategy library presets a variety of response plans according to different dangerous situations, including information such as the activation order, brightness settings, and working hours of lamps in different areas. For example, in a low-risk state, only emergency lighting fixtures in certain key areas are activated; in a high-risk state, all emergency lighting equipment in the entire building is activated. By selecting the corresponding basic lighting strategy, a preliminary first lighting plan is generated. Perform spatial distribution analysis on the lighting fixtures in the lighting control system to obtain the location information of each lighting fixture. and coverage area ,in Indicates The coordinates of the lamps, Indicates its lighting range. Through analysis, the geographical location of the lamps is understood, and it is determined whether there are blind spots or overlapping areas in the lighting coverage. Based on the location information and coverage area information, the first lighting plan is optimized to generate a more reasonable second lighting plan. The optimization process includes considering how to effectively activate lamps to avoid unnecessary energy consumption, ensuring that lamps in dangerous areas are activated first, and reasonably allocating them according to the coverage range of the lamps. For example, if the coverage area of ​​two adjacent lamps is and If there is a large overlap, choose to activate only one lamp to reduce power consumption. At the same time, for lamps that cover fewer critical areas or are far away from dangerous sources, appropriately reduce their activation priority or adjust their brightness to improve the overall efficiency of the emergency lighting system. Evaluate the power of each lighting fixture to be activated in the second lighting scheme to ensure that the lamp has sufficient power to maintain operation when activated. The power evaluation formula is expressed as: ;

[0107] in, Indicates The power status of each lamp, is the remaining capacity of the lamp battery, is the power of the lamp, is the estimated duration of emergency lighting. The formula is used to determine whether each lamp has enough power to support the expected lighting time. According to the results of the power status assessment, the second lighting plan is adjusted to generate an emergency lighting plan. If some lamps are low on power, they are replaced or their brightness is reduced to save power. In some cases, the activation order of other lamps is adjusted to ensure that the emergency lighting system can maintain sufficient lighting levels even when power is limited. The final generated emergency lighting plan includes information on all lamps to be activated, including the activation time, brightness setting and power status of each lamp.

[0108] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0109] Analyze the emergency lighting solution to obtain a list of lighting fixtures to be activated and control parameters of each lighting fixture to be activated;

[0110] Based on the list of lighting fixtures to be activated, the current communication state of the lighting control system is detected to obtain a list of candidate communication modes;

[0111] According to the candidate communication mode list, the lighting fixtures to be activated are grouped to obtain a wired communication group and a wireless communication group;

[0112] Performing address coding on the lighting fixtures in the wired communication group to obtain a wired communication instruction sequence, and performing channel allocation on the lighting fixtures in the wireless communication group to obtain a wireless communication instruction sequence;

[0113] The wired communication instruction sequence and the wireless communication instruction sequence are combined to obtain a target control instruction set, and the target control instruction set is encrypted to obtain a control instruction;

[0114] Based on the preset communication protocol, the control instruction is sent to the corresponding lighting fixture to be activated.

[0115] Specifically, the emergency lighting scheme is parsed to obtain a list of lighting fixtures to be activated and the control parameters of each fixture. The emergency lighting scheme contains specific information about the lighting fixtures to be activated, such as the number, location, power setting, brightness requirement, etc. of the fixtures. By parsing the scheme, a list of lighting fixtures to be activated is generated, listing all the fixtures that are about to enter emergency mode. For example, suppose there is The lamps to be activated are recorded as , and analyze the control parameters of each lamp, such as power ,brightness And the running time Etc. These control parameters ensure that each lamp can work according to the preset requirements under emergency conditions. After generating a list of lamps to be activated, detect the current communication status of the lighting control system to determine the appropriate communication method. Modern lighting control systems usually support both wired and wireless communication methods. Therefore, it is necessary to detect the current communication network and evaluate the availability of wired communication and the status of the wireless network, such as signal strength, interference level, and network delay. Through detection, a list of candidate communication methods is generated, which includes all available communication methods. According to the list of candidate communication methods, the lighting fixtures to be activated are grouped to form wired communication groups and wireless communication groups. Assume that some of the lamps in the lighting system Support wired communication, another part of the lamps Relying on wireless communication, the lamps are assigned to wired communication groups and wireless communication groups respectively. The lamps in the wired communication group transmit data through network cables, while the lamps in the wireless communication group communicate with the control unit through wireless channels. Address encoding is performed on the lighting fixtures in the wired communication group. Each lighting fixture has a unique address in the wired network, which is generated based on the IP address or other network address allocation rules. By address encoding the lamps, it is ensured that the control instructions can be accurately delivered to each specified lamp. Assuming that the lamp The address is , lamps The address is , lamps The address is , and these addresses are associated with the corresponding control parameters and The wireless communication groups are bound together to form a wired communication instruction sequence, which contains the information required to send control instructions to each wired lamp. For lighting fixtures in the wireless communication group, an appropriate wireless communication channel is allocated to each lamp to avoid interference and signal congestion. Assuming the system is a lamp Assign channel These channels ensure that each lamp can receive control instructions through stable wireless communication. and Bind together to form a wireless communication instruction sequence. Combine the wired communication instruction sequence and the wireless communication instruction sequence to obtain a target control instruction set, which contains the control instructions of all lamps to be activated. In order to ensure the security of data transmission, the target control instruction set is encrypted to prevent the data from being intercepted or tampered with during transmission, ensuring that the control instructions can be safely and accurately delivered to each lamp. Encryption uses a symmetric encryption algorithm, such as AES (Advanced Encryption Standard), or an asymmetric encryption algorithm such as RSA. The encryption formula is expressed as: ;

[0116] in, Indicates the encrypted control instructions. is the encryption function, is the encryption key, It is the original control instruction. Through this process, the security and integrity of the instruction during transmission are ensured. Based on the preset communication protocol, the encrypted control instruction is sent to the corresponding lighting fixture to be activated. The instructions of the wired communication group are sent through the wired network, while the instructions of the wireless communication group are sent to the lamps through the allocated wireless channel. Different communication protocols have different transmission methods. The wired protocol includes TCP / IP, and the wireless protocol includes Wi-Fi, ZigBee or LoRa.

[0117] In a specific embodiment, the process of executing step S6 may specifically include the following steps:

[0118] Decrypting the received control instructions to obtain decrypted control parameters;

[0119] Based on the decrypted control parameters, the power supply system of the lighting fixture is switched from the 48V centralized power supply interface to the single-point power supply system with built-in batteries;

[0120] According to the brightness setting in the decrypted control parameters, the driving current of the LED light source is adjusted to obtain the target brightness output;

[0121] Based on the power setting in the decrypted control parameters, the power control circuit of the lighting fixture is adjusted to switch the power from the normal working mode power value to the emergency mode power value;

[0122] Detect the power level of the built-in battery of the lighting fixture to obtain the current power status;

[0123] According to the current power status and the preset emergency lighting duration requirements, the brightness of the LED light source is dynamically adjusted to obtain the optimized brightness output;

[0124] Based on the optimized brightness output, the cooling system of the lighting fixture is adjusted to obtain the emergency lighting working status;

[0125] Provide status feedback on the working status of emergency lighting and send the status information to the control unit through the preset communication protocol.

[0126] Specifically, the received control instruction is decrypted to obtain the decrypted control parameters. The corresponding decryption algorithm is called to restore the encrypted data to the original control parameters by using the corresponding key. Assuming that the control instruction is encrypted by AES (Advanced Encryption Standard), the decryption process is expressed as: ;

[0127] in, is the decrypted control parameter, is the decryption function, For the decryption key, is an encrypted control instruction. Through the decryption process, the specific control parameters required for each lamp are obtained, including key information such as brightness setting, power mode, and emergency duration. After successful decryption, the power supply system of the lighting fixture is switched based on the decrypted control parameters. When a problem with the external power supply is detected, or the lamp is required to switch to emergency mode according to the instruction, the 48V centralized power supply is switched to the built-in battery power supply. The power supply switching operation is completed by controlling the power switching module inside the lamp to ensure that the lighting fixture can continue to operate in the event of a power outage or other emergency. According to the brightness setting in the decrypted control parameters, the driving current of the LED light source is adjusted to achieve the target brightness output. The brightness of the LED is proportional to its driving current, and the driving current is adjusted to accurately control the brightness of the lamp. Assuming the brightness With drive current The relationship is: ;

[0128] in, is a constant, which is related to the luminous efficiency of the LED light source. is the drive current. By adjusting , set the target brightness output of the LED light source according to the decrypted control parameters. For example, if the decrypted parameters require the lamp to output 50% brightness, the system will adjust the drive current accordingly to ensure that the actual output light intensity matches the expected one. Based on the power setting in the decrypted control parameters, adjust the power control circuit of the lighting fixture. The power in emergency mode is usually lower than the normal operating mode to reduce energy consumption and extend battery life. Assume that the power of the lamp in normal mode is , the power in emergency mode is , then according to the control parameters, the power is Switch to The power regulation is achieved by changing the input voltage or current to ensure that the lamp can balance the brightness and power consumption in an emergency state. In order to optimize the emergency working state of the lamp, the built-in battery of the lighting fixture is tested to obtain the current remaining battery power. , and adjust the working strategy of the lamp according to the power status. For example, if the battery is low, the brightness of the lamp is reduced or the working time is reduced to ensure that the lighting time can be maximized in an emergency. The results of the power detection are used for subsequent dynamic brightness adjustment. According to the current power status and the preset emergency lighting duration requirements, the brightness of the LED light source is dynamically adjusted to ensure that the battery power is sufficient to maintain lighting until the end of the emergency state. Send to the central control system through wired or wireless communication protocols to ensure that the control unit can monitor the status of each lamp in real time. If a lamp fails or the power is too low, the control unit takes timely measures to adjust the working status of other lamps to ensure the overall stability of the emergency lighting.

[0129] The above describes the lighting control method in an emergency situation in the embodiment of the present invention. The following describes the lighting control system in an emergency situation in the embodiment of the present invention. Figure 2 , an embodiment of the lighting control system in an emergency situation in an embodiment of the present invention includes:

[0130] A configuration module, used to configure a lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit;

[0131] A monitoring module, used for monitoring data of environmental sensors through a control unit;

[0132] An analysis module, used for the control unit to analyze the dangerous situation and obtain the danger level when the data of the environmental sensor indicates the existence of a dangerous situation;

[0133] A generating module, configured to generate an emergency lighting plan by the control unit according to the danger level, wherein the emergency lighting plan includes information of the lighting fixtures to be activated;

[0134] A sending module, used for sending control instructions from the control unit to the lighting fixtures to be activated based on the emergency lighting solution;

[0135] The switching module is used to switch the lighting fixture to be activated to the emergency lighting state in response to the control instruction.

[0136] Through the collaboration of the above components, the multi-branch fusion spatiotemporal graph convolutional network is used to process the environmental state spatiotemporal graph, which realizes the deep fusion analysis of multi-dimensional data such as smoke, temperature, carbon monoxide and electricity, and improves the accuracy and comprehensiveness of dangerous situation judgment. The dynamic selection of the optimal data transmission path, combined with the dual communication interfaces of wired and wireless, enhances the communication reliability and flexibility of the system in a complex network environment. The emergency lighting scheme is automatically generated and optimized based on the hazard level, and the spatial distribution and power status of lighting fixtures are considered to achieve efficient utilization and precise control of lighting resources. By setting and encrypting the control parameters of each lighting fixture to be activated separately, the security and pertinence of the system are improved, and it can adapt to the lighting needs of different areas. Intelligent switching and adaptive adjustment of lighting fixtures are realized, including power supply system switching, dynamic adjustment of brightness and power, and corresponding adjustment of the heat dissipation system, ensuring the durability and stability of emergency lighting. Through the state feedback mechanism, the control unit can grasp the working status of each lighting fixture in real time.

[0137] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0138] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or 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 method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0139] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A lighting control method in an emergency situation, characterized in that: The method comprises: configuring a lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit; The data of the environmental sensor is monitored by the control unit; specifically including: collecting data from the environmental sensor to obtain original environmental data, and classifying the original environmental data to obtain smoke concentration data, temperature change rate data, carbon monoxide concentration data and power supply status data; filtering the smoke concentration data to obtain filtered smoke concentration data; calculating the temperature change rate data to obtain the temperature change value per unit time; standardizing the carbon monoxide concentration data to obtain standardized carbon monoxide concentration data; parsing the power supply status data to obtain voltage and current values; parallel detection of the wired network and the wireless network in the lighting control system to obtain wired network status parameters and wireless network status parameters; calculating the connection quality score of the wired network based on the wired network status parameters to obtain the wired network score, and calculating the signal strength of the wireless network based on the wireless network status parameters. degree and stability to obtain a wireless network score; compare and analyze the wired network score and the wireless network score to obtain a current communication mode, and perform network topology analysis based on the current communication mode to obtain a network structure model; perform path planning on the network structure model to obtain multiple candidate data transmission paths, and evaluate the multiple candidate data transmission paths based on a preset path evaluation algorithm to obtain a transmission efficiency index of each candidate data transmission path; sort the multiple candidate data transmission paths according to the transmission efficiency index to obtain a sorted path list, and select the path with the highest transmission efficiency index from the sorted path list as the optimal data transmission path; based on the optimal data transmission path, periodically transmit the filtered smoke concentration data, the temperature change value per unit time, the standardized carbon monoxide concentration data, the voltage value and the current value to obtain the data of the environmental sensor; When the data of the environmental sensor indicates the existence of a dangerous situation, the control unit performs a dangerous situation analysis to obtain a dangerous level; According to the danger level, the control unit generates an emergency lighting plan, wherein the emergency lighting plan includes information of lighting fixtures to be activated; Based on the emergency lighting solution, the control unit sends a control instruction to the lighting fixture to be activated; In response to the control instruction, the lighting fixture to be activated switches to an emergency lighting state.

2. The lighting control method in emergency situations according to claim 1, characterized in that: The configuration lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit, comprises: Allocating addresses to a plurality of lighting fixtures to obtain a unique identification address for each lighting fixture, and configuring a power supply system for the plurality of lighting fixtures so that each lighting fixture has a 48V centralized power supply interface and a single-point power supply system with a built-in battery; Performing light source configuration on the plurality of lighting fixtures so that each lighting fixture has an adjustable LED light source, and performing communication interface configuration on the control unit so that the control unit has dual communication interfaces, wired and wireless; Based on the user interface of the control unit, the working mode power of each lighting fixture is set to obtain a normal working mode power value and an emergency mode power value, and based on the user interface of the control unit, the lighting fixture is designated in emergency mode to obtain an emergency lighting fixture list; Based on the user interface of the control unit, the emergency mode brightness is set for the lamps in the emergency lighting fixture list to obtain the emergency mode brightness value of each emergency lighting fixture, and the lighting control system is self-checked to verify the functional status of all components to obtain a configured lighting control system.

3. The lighting control method in emergency situations according to claim 1, characterized in that: When the data of the environmental sensor indicates that a dangerous situation exists, the control unit performs a dangerous situation analysis to obtain a dangerous level, including: Reconstructing the filtered smoke concentration data, the temperature change value per unit time, the standardized carbon monoxide concentration data, the voltage value and the current value in the data of the environmental sensor to obtain multi-dimensional time series data; Constructing a spatiotemporal graph based on the multidimensional time series data to obtain a spatiotemporal graph of the environmental state; Performing multi-branch fusion spatiotemporal graph convolutional network processing on the environmental state spatiotemporal graph to obtain a dangerous situation feature vector, wherein the multi-branch fusion spatiotemporal graph convolutional network includes a smoke branch, a temperature branch, a carbon monoxide branch, and a power branch; the multi-branch fusion spatiotemporal graph convolutional network is deployed in the control unit; In the smoke branch, a time convolution operation is performed on the environment state spatiotemporal graph to obtain a smoke time feature, and a graph convolution operation is performed on the smoke time feature to obtain a smoke space feature, and then the smoke time feature and the smoke space feature are feature fused to obtain a smoke hazard feature; the smoke branch includes: three layers of time convolution layers, each layer uses 64 3x1 convolution kernels and adopts a ReLU activation function; two layers of graph convolution layers, each layer uses 32 Chebyshev polynomial kernels and adopts an ELU activation function; a feature fusion layer, which uses a gated linear unit for feature fusion; In the temperature branch, a graph convolution operation is performed on the environmental state spatiotemporal graph to obtain a temperature spatial feature, and a time convolution operation is performed on the temperature spatial feature to obtain a temperature time feature, and then the temperature spatial feature and the temperature time feature are feature fused to obtain a temperature hazard feature; the temperature branch includes: two graph convolution layers, each layer uses 48 Diffusion convolution kernels and adopts a LeakyReLU activation function; three time convolution layers, each layer uses 56 2x1 convolution kernels and adopts a GELU activation function; a feature fusion layer, which uses a self-attention mechanism for feature fusion; In the carbon monoxide branch, the environmental state spatiotemporal graph is subjected to parallel time convolution and graph convolution operations to obtain carbon monoxide time features and carbon monoxide spatial features respectively, and then the carbon monoxide time features and the carbon monoxide spatial features are fused by an attention mechanism to obtain carbon monoxide hazard features; the carbon monoxide branch comprises: two parallel sub-branches, the time convolution sub-branch comprises three 1D convolution layers, each layer uses 40 convolution kernels, and adopts a Swish activation function; the graph convolution sub-branch comprises two graph convolution layers, each layer uses 36 spectral convolution kernels, and adopts a Mish activation function; a multi-head attention layer for feature fusion, comprising 4 attention heads; In the power branch, the environment state spatiotemporal graph is processed by a graph attention network to obtain power space features, and the power space features are processed by a gated recurrent unit to obtain power time features, and then the power space features and the power time features are adaptively fused to obtain power hazard features; the power branch includes: three layers of graph attention layers, each layer contains 6 attention heads, using residual connections and layer normalization; two layers of gated recurrent units, each layer contains 128 hidden units, using Tanh activation function; an adaptive fusion layer, using a dynamic weight generation network for feature fusion; Inputting the smoke hazard feature, the temperature hazard feature, the carbon monoxide hazard feature and the power hazard feature into a multilayer perceptron to obtain a fused hazard feature; the multilayer perceptron includes three fully connected layers, the number of hidden layer units is 256, 128 and 64 respectively, each layer is followed by a batch normalization and a Dropout layer, and the activation function adopts ReLU; The fused danger feature is subjected to softmax classification to obtain a danger level probability distribution. When the danger level probability distribution indicates the existence of a dangerous situation, a weighted voting analysis is performed based on the danger level probability distribution to obtain the danger level.

4. The lighting control method in emergency situations according to claim 3, characterized in that: In the carbon monoxide branch, the environmental state spatiotemporal graph is subjected to parallel time convolution and graph convolution operations to obtain carbon monoxide time features and carbon monoxide spatial features respectively, and then the carbon monoxide time features and the carbon monoxide spatial features are fused by an attention mechanism to obtain carbon monoxide hazard features, including: Slicing the environmental state space-time diagram in the time dimension to obtain carbon monoxide concentration time series data; Input the carbon monoxide concentration time series data into the first 1D convolution layer, perform a convolution operation using 40 convolution kernels to obtain a first-layer time feature map, and perform a Swish activation function on the first-layer time feature map to obtain a first-layer activated time feature map; The activated time feature graph of the first layer is sequentially input into the second and third 1D convolution layers, 40 convolution kernels are used for convolution operation in each layer and the Swish activation function is applied to obtain the carbon monoxide time feature, and the environmental state spatiotemporal graph is sliced ​​in the spatial dimension to obtain the carbon monoxide concentration spatial distribution data; Inputting the carbon monoxide concentration spatial distribution data into the first graph convolution layer, performing a graph convolution operation using 36 spectral convolution kernels to obtain a first-layer spatial feature map, and performing a Mish activation function on the first-layer spatial feature map to obtain a first-layer activated spatial feature map; The activated spatial feature map of the first layer is input into the second graph convolution layer, 36 spectral convolution kernels are used to perform graph convolution operation and Mish activation function is applied to obtain the spatial feature of carbon monoxide, and the temporal feature of carbon monoxide and the spatial feature of carbon monoxide are concatenated to obtain the spatiotemporal joint feature; The spatiotemporal joint features are subjected to multi-head attention processing, and four attention heads are used to calculate the self-attention weights to obtain the carbon monoxide hazard features.

5. The lighting control method in emergency situations according to claim 1, characterized in that: The control unit generates an emergency lighting plan according to the danger level, wherein the emergency lighting plan includes information of lighting fixtures to be activated, including: Performing threshold judgment on the danger level to obtain a danger degree classification result; Based on the hazard level classification result, a corresponding basic lighting strategy is selected from a preset emergency lighting strategy library to obtain a first lighting plan; Performing spatial distribution analysis on the lighting fixtures in the lighting control system to obtain location information and coverage area information of the lighting fixtures; Based on the location information and the coverage area information, the first lighting scheme is optimized to obtain a second lighting scheme; Performing power evaluation on each lighting fixture to be activated in the second lighting scheme to obtain power status information; The second lighting scheme is adjusted according to the power status information to obtain an emergency lighting scheme, wherein the emergency lighting scheme includes information about the lighting fixtures to be activated.

6. The lighting control method in emergency situations according to claim 1, characterized in that: The step of sending a control instruction to the lighting fixture to be activated by the control unit based on the emergency lighting scheme includes: Parsing the emergency lighting scheme to obtain a list of lighting fixtures to be activated and control parameters of each lighting fixture to be activated; Based on the list of lighting fixtures to be activated, detecting the current communication state of the lighting control system to obtain a list of candidate communication modes; According to the candidate communication mode list, the lighting fixtures to be activated are grouped to obtain a wired communication group and a wireless communication group; Performing address coding on the lighting fixtures in the wired communication group to obtain a wired communication instruction sequence, and performing channel allocation on the lighting fixtures in the wireless communication group to obtain a wireless communication instruction sequence; The wired communication instruction sequence and the wireless communication instruction sequence are combined to obtain a target control instruction set, and the target control instruction set is encrypted to obtain a control instruction; Based on a preset communication protocol, the control instruction is sent to the corresponding lighting fixture to be activated.

7. The lighting control method in emergency situations according to claim 6, characterized in that: In response to the control instruction, the lighting fixture to be activated switches to the emergency lighting state, comprising: Decrypting the received control instruction to obtain decrypted control parameters; Based on the decrypted control parameters, the power supply system of the lighting fixture is switched from a 48V centralized power supply interface to a single-point power supply system with a built-in battery; According to the brightness setting in the decrypted control parameters, the driving current of the LED light source is adjusted to obtain the target brightness output; Based on the power setting in the decrypted control parameter, adjusting the power control circuit of the lighting fixture to switch the power from the normal working mode power value to the emergency mode power value; Detect the power level of the built-in battery of the lighting fixture to obtain the current power status; According to the current power state and the preset emergency lighting duration requirement, the brightness of the LED light source is dynamically adjusted to obtain an optimized brightness output; Based on the optimized brightness output, the heat dissipation system of the lighting fixture is adjusted to obtain the emergency lighting working state; The emergency lighting operating status is fed back and the status information is sent to the control unit via a preset communication protocol.

8. A lighting control system for emergency situations, characterized in that: The system is used to perform the lighting control method in an emergency situation according to any one of claims 1 to 7, the system comprising: A configuration module, used to configure a lighting control system, the lighting control system comprising a plurality of lighting fixtures, at least one environmental sensor and a control unit; A monitoring module, used for monitoring the data of the environmental sensor through the control unit; An analysis module, configured to cause the control unit to analyze the dangerous situation and obtain a dangerous level when the data of the environmental sensor indicates that a dangerous situation exists; A generating module, configured to generate an emergency lighting plan by the control unit according to the danger level, wherein the emergency lighting plan includes information of lighting fixtures to be activated; A sending module, configured to send a control instruction from the control unit to the lighting fixture to be activated based on the emergency lighting solution; A switching module is used to switch the lighting fixture to be activated to an emergency lighting state in response to the control instruction.

Images