Linkage illumination control system and method based on Internet of Things

Abstract

The invention discloses a linkage illumination control system and method based on the Internet of Things, and belongs to the technical field of the Internet of Things. The system monitors the natural light brightness, personnel and vehicle activities, forklift positions and ambient temperature of each area of a warehouse, preprocesses the collected data, and transmits the data to a local server and a cloud server; performing feature extraction, and combining the extracted features into a comprehensive feature vector; using a multi-layer perceptron to train a machine learning model, designing the structure of the multi-layer perceptron, performing forward propagation, loss calculation, back propagation and weight updating until the model converges, and outputting a lighting control instruction; constructing an energy consumption model, defining a multi-objective optimization problem, and solving the multi-objective optimization problem through a particle swarm optimization algorithm to obtain an optimal illumination control instruction; when it is detected that the illumination intensity of a certain area is lower than a threshold value, a fault alarm is automatically generated; when the brightness of the natural light is detected to be insufficient, increasing the brightness of the surrounding light source; and carrying out power failure processing.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a linkage lighting control system and method based on the Internet of Things. Background Art

[0002] With the rapid development of modern industry, warehouses are an important part of logistics and storage, and the management of their lighting systems is particularly important. Traditional warehouse lighting systems usually use fixed time period control or manual control, which not only wastes energy but also cannot respond to changes in the warehouse environment in real time. This inefficient lighting management method can no longer meet the needs of modern smart warehouses.

[0003] Since the lighting brightness cannot be adjusted according to actual needs, traditional systems often maintain high brightness lighting during the day or in unmanned areas, resulting in a large amount of energy waste. Traditional systems cannot automatically detect light source failures and require regular manual inspection and maintenance, which increases labor costs and maintenance difficulties. The lighting cannot be dynamically adjusted according to the activities of people and vehicles in the warehouse, affecting the efficiency and safety of warehouse operations. Summary of the invention

[0004] The purpose of the present invention is to provide a linkage lighting control system and method based on the Internet of Things to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: A linkage lighting control method based on the Internet of Things, the method comprising the following steps: S100, using sensors to monitor the natural light brightness in various areas of the warehouse, monitor personnel and vehicle activities, track the location of forklifts in real time, monitor ambient temperature, pre-process the collected data, and transmit it to local servers and cloud servers; S200, obtaining preprocessed data from a cloud server, performing feature extraction, wherein the features include light intensity features, motion features, position features, and temperature features, and combining the extracted features into a comprehensive feature vector; using a multi-layer perceptron to train a machine learning model, designing a structure of the multi-layer perceptron, performing forward propagation, loss calculation, back propagation, and weight update until the model converges, and outputting a lighting control instruction; S300, build an energy consumption model to calculate the energy consumption under different lighting settings; define a multi-objective optimization problem to balance lighting demand and energy consumption; solve the multi-objective optimization problem through a particle swarm optimization algorithm, combine the lighting control instructions to obtain the optimal lighting control instructions, apply them to the lighting system of the warehouse, and adjust the brightness and status of each lamp; S400. Set the minimum threshold of natural light brightness. When the system detects that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm. When the system detects that the natural light brightness is insufficient, the system increases the brightness of the surrounding light sources according to the preset strategy. It also handles power failures and automatically switches to backup power.

[0006] According to step S100, light sensors are installed in various areas of the warehouse to monitor the brightness of natural light, motion sensors are installed in main passages and work areas to monitor the activities of personnel and forklifts, position sensors are installed in forklifts and other transportation equipment to track positions in real time, temperature sensors are installed in various areas of the warehouse to monitor ambient temperature, and the data sampling frequency of each sensor is set; Use median filtering to remove outliers and noise in sensor data, perform outlier detection on data beyond the normal range, use interpolation to process missing data points, and normalize or standardize sensor data to unify the data dimension; transmit the preprocessed sensor data to the local server for storage through the IoT gateway; synchronously transmit the preprocessed data to the cloud server; during data transmission, use SSL / TLS to ensure data security, encrypt the stored data, and set access control policies.

[0007] According to step S200, when extracting the light intensity feature, it is necessary to obtain the current light intensity and the daily average light change, wherein the current light intensity directly uses the light intensity value of each sampling point , where n is the total number of sampling points, They are the light intensity values ​​from sampling point 1 to sampling point n respectively; to obtain the daily average light change, it is necessary to calculate the rate of change of light intensity in each time period , let each time period be , the formula is as follows: ; Calculate the average rate of change , as the daily average light change, the formula is as follows: ; When extracting motion features, it is necessary to obtain the activity frequency and intensity of people and vehicles. The number of activities of people and vehicles in each time period is counted. The activity detection data is in binary form, 0 means no activity, 1 means activity, and the total number of activities is for: ; in, is the number of activities at the i-th sampling point; activity intensity It is measured by calculating the duration of the activity in each time period, and the activity duration is , the total time is T, and the formula for activity intensity is as follows: ; When extracting position features, it is necessary to obtain the forklift position and position change speed, where the forklift position is represented as a three-dimensional coordinate , let the forklift position of the i-th sampling point be , then the forklift position It is expressed as: ; Calculate the position change speed in each time period, the position change of the x-axis is , the position of the y-axis changes to , the position change of z axis is , position change speed The calculation formula is as follows: ; Calculate average speed , as the position change speed, the formula is as follows: ; When extracting temperature features, the current temperature and daily temperature change are required. The current temperature directly uses the temperature value of each sampling point. ; Calculate the rate of change of temperature in each time period , the formula is as follows: ; Calculate the average rate of change , as the daily temperature change, the formula is as follows: ; Combine the features extracted at sampling point i into independent comprehensive feature vectors at different sampling points , expressed as follows: ; The independent comprehensive feature vectors of all different sampling points Combine to form a comprehensive feature vector .

[0008] According to step S200, a multilayer perceptron model is designed, which includes an input layer, two hidden layers and an output layer, wherein the input layer includes n neurons corresponding to the dimension of the comprehensive feature vector, and the first hidden layer includes neurons, and the second hidden layer contains The activation function of the first hidden layer and the second hidden layer is ReLU. The output layer contains m neurons, corresponding to the dimension of the lighting control instructions, and the activation function is Softmax.

[0009] According to step S200, forward propagation is performed, and the input feature vector is a comprehensive feature vector , with a dimension of n, the output of the first hidden layer is ,in, is the weight matrix from the input layer to the first hidden layer, with a dimension of , is the bias vector of the first hidden layer, with dimension ; The output of the second hidden layer is ,in, is the weight matrix from the first hidden layer to the second hidden layer, with a dimension of , is the bias vector of the second hidden layer, with dimension ; The output of the output layer is y, ,in, is the weight matrix from the second hidden layer to the output layer, with a dimension of , is the bias vector of the output layer, with dimension ; Assume the target output is , using mean square error as the loss function for calculation, the formula is as follows: ; in, is the loss function, is the ith predicted output, is the i-th true output.

[0010] According to step S200, the gradient of the loss function to the weights and biases of each layer is calculated by the back propagation algorithm; the gradient calculation formula from the output layer to the second hidden layer is as follows: ; ; in, is the loss function for The gradient of is the loss function for The gradient of is the transpose of the output vector of the second hidden layer; The gradient calculation formula from the second hidden layer to the first hidden layer is as follows: ; in, is the error term of the second hidden layer, yes The transpose of is the derivative of the ReLU activation function of the second hidden layer, represents element-wise product; ; ; in, is the loss function for The gradient of is the loss function for The gradient of is the transpose of the output vector of the first hidden layer; The gradient calculation formula from the first hidden layer to the input layer is as follows: ; in, is the error term of the first hidden layer, yes The transpose of is the derivative of the ReLU activation function of the first hidden layer; ; ; in, is the loss function for The gradient of is the loss function for The gradient of is the transpose of the input vector.

[0011] According to step S200, the weight updating process is as follows: using the gradient descent method, updating the weight and bias of the output layer according to the calculated gradient, updating the weight and bias of the second hidden layer, and updating the weight and bias of the first hidden layer, the step size of the weight update is determined by the learning rate, and the learning rate is a pre-set hyperparameter; The process of model training is as follows: initialize the weights and biases of all layers to random values; divide all training samples into several batches, and perform forward propagation, loss calculation and back propagation once for each batch; average the gradients of each batch to update the weights and biases; repeat the above steps until the loss function converges or reaches the preset number of iterations; at the end of each iteration cycle, evaluate the performance of the model on the validation set to monitor the training process and prevent overfitting; The trained model performs forward propagation based on the real-time input sensor data to generate corresponding lighting control instructions. The output lighting control instructions are used to adjust the warehouse's lighting system.

[0012] According to step S300, input variables of the lighting system are defined, wherein the input variables of the lighting system include the brightness setting of each lamp, the natural light brightness of each area, and the ambient temperature; an energy consumption model is constructed, including the following factors: the relationship between the power consumption of each lamp and its brightness and the total energy consumption of each area, wherein the total energy consumption of each area is the sum of the energy consumption of all lamps; The brightness setting of each lamp indicates the light output intensity of the lamp, which can be a number between 0 and the maximum brightness value; the natural light brightness of each area indicates the light intensity from natural light sources in a specific area, which is obtained through real-time monitoring by light sensors; the ambient temperature in the warehouse affects the efficiency and life of the lamps, which is obtained through real-time monitoring by temperature sensors.

[0013] Define a multi-objective optimization problem. For the first objective, ensure that the light intensity of each area is greater than or equal to the minimum light requirement. For the second objective, optimize the brightness setting of each lamp to minimize the total energy consumption. The process of using the particle swarm optimization algorithm is as follows: generate a particle swarm, each particle represents a combination of lamp brightness settings; each particle updates the brightness setting and speed based on its own experience and the experience of its neighbors; calculate the fitness of each particle, which is the satisfaction of energy consumption and lighting requirements; update the individual optimal position of each particle and the global optimal position of the group; repeat the above steps until the preset number of iterations is reached; the position of the optimal particle is combined with the lighting control instruction as the optimal lighting control instruction and applied to the warehouse lighting system to adjust the brightness and status of each lamp.

[0014] According to step S400, a minimum threshold of natural light brightness in each area is defined, and a light sensor is used to monitor the natural light brightness in each area in real time. When the natural light brightness in a certain area is lower than the threshold, the system automatically generates a fault alarm and sends an alarm message to the staff, prompting them to check whether the light source is damaged or blocked; The system continuously monitors the brightness of natural light. When it detects that the brightness of natural light in a certain area is insufficient, it automatically increases the brightness of the lamps around the area according to the preset strategy to ensure that the total light intensity meets the requirements. The system monitors the power status of the warehouse in real time. When a power failure is detected, the system automatically switches to the backup power supply. The switching process is seamless to avoid lighting interruptions. A power failure alarm is sent to the staff, prompting them to check the cause of the main power failure.

[0015] A linkage lighting control system based on the Internet of Things includes the following steps: Data collection and preprocessing module: including: data collection unit and data preprocessing unit; the data collection unit uses sensors to monitor the natural light brightness in various areas of the warehouse, monitor personnel and vehicle activities, track the location of forklifts in real time, and monitor ambient temperature; the data preprocessing unit preprocesses the collected data and transmits it to the local server and cloud server; Feature extraction and model training module: including: feature extraction unit, multi-layer perceptron design unit and model training unit; wherein, the feature extraction unit is responsible for obtaining pre-processed data from the cloud server and extracting features, wherein the features include light intensity features, motion features, position features and temperature features, and combining the extracted features into a comprehensive feature vector; the multi-layer perceptron design unit uses the multi-layer perceptron to train the machine learning model and designs the structure of the multi-layer perceptron; the model training unit is responsible for forward propagation, loss calculation, back propagation and weight update until the model converges and outputs the lighting control instructions; Energy consumption optimization module: including: energy consumption model construction unit, multi-objective optimization problem definition unit and particle swarm optimization algorithm unit; the energy consumption model construction unit is responsible for constructing an energy consumption model to calculate the energy consumption under different lighting settings, the multi-objective optimization problem definition unit is responsible for defining multi-objective optimization problems to balance lighting demand and energy consumption, and the particle swarm optimization algorithm unit solves the multi-objective optimization problem through the particle swarm optimization algorithm, combines the lighting control instructions to obtain the optimal lighting control instructions, and applies them to the lighting system of the warehouse to adjust the brightness and status of each lamp; Fault detection and processing module: including: fault alarm unit, light source adjustment unit and power fault processing unit; among them, the fault alarm unit is responsible for setting the minimum threshold of natural light brightness. When it is detected that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm. When the light source adjustment unit detects that the natural light brightness is insufficient, the system increases the brightness of the surrounding light sources according to the preset strategy. The power fault processing unit is responsible for power fault processing and automatically switching to the backup power supply.

[0016] Compared with the prior art, the beneficial effects achieved by the present invention are: By constructing an energy consumption model and using a particle swarm optimization algorithm, the present invention can optimize the brightness setting of each lamp while meeting lighting needs, achieve the lowest energy consumption, and significantly reduce energy costs.

[0017] The present invention sets a minimum threshold for natural light brightness. When it detects that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm and promptly notifies the staff to conduct an inspection. At the same time, when natural light is insufficient, the system automatically increases the brightness of the surrounding light sources to ensure lighting needs.

[0018] The present invention has a power failure processing function. When a power outage is detected, the system will automatically switch to a backup power supply to ensure the continuous operation of the lighting system and avoid operation interruption. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 It is a schematic diagram of the steps of a linkage lighting control system and method based on the Internet of Things of the present invention; Figure 2 It is a system structure diagram of a linkage lighting control system and method based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] See also Figure 1 and Figure 2 , the present invention provides a technical solution: According to one embodiment of the present invention, Figure 1 A step diagram of a linkage lighting control system and method based on the Internet of Things is shown in a schematic diagram, a linkage lighting control method based on the Internet of Things, the method comprising the following steps: S100, using sensors to monitor the natural light brightness in various areas of the warehouse, monitor personnel and vehicle activities, track the location of forklifts in real time, monitor ambient temperature, pre-process the collected data, and transmit it to local servers and cloud servers; S200, obtaining preprocessed data from a cloud server, performing feature extraction, wherein the features include light intensity features, motion features, position features, and temperature features, and combining the extracted features into a comprehensive feature vector; using a multi-layer perceptron to train a machine learning model, designing a structure of the multi-layer perceptron, performing forward propagation, loss calculation, back propagation, and weight update until the model converges, and outputting a lighting control instruction; S300, build an energy consumption model to calculate the energy consumption under different lighting settings; define a multi-objective optimization problem to balance lighting demand and energy consumption; solve the multi-objective optimization problem through a particle swarm optimization algorithm, combine the lighting control instructions to obtain the optimal lighting control instructions, apply them to the lighting system of the warehouse, and adjust the brightness and status of each lamp; S400. Set the minimum threshold of natural light brightness. When the system detects that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm. When the system detects that the natural light brightness is insufficient, the system increases the brightness of the surrounding light sources according to the preset strategy. It also handles power failures and automatically switches to backup power.

[0022] According to step S100, light sensors are installed in various areas of the warehouse to monitor the brightness of natural light, motion sensors are installed in main passages and work areas to monitor the activities of personnel and forklifts, position sensors are installed in forklifts and other transportation equipment to track positions in real time, temperature sensors are installed in various areas of the warehouse to monitor ambient temperature, and the data sampling frequency of each sensor is set; Use median filtering to remove outliers and noise in sensor data, perform outlier detection on data beyond the normal range, use interpolation to process missing data points, and normalize or standardize sensor data to unify the data dimension; transmit the preprocessed sensor data to the local server for storage through the IoT gateway; synchronously transmit the preprocessed data to the cloud server; during data transmission, use SSL / TLS to ensure data security, encrypt the stored data, and set access control policies.

[0023] According to step S200, when extracting the light intensity feature, it is necessary to obtain the current light intensity and the daily average light change, wherein the current light intensity directly uses the light intensity value of each sampling point , where the total number of sampling points is 5; to obtain the average daily light change, it is necessary to calculate the rate of change of light intensity in each time period , let each time period be Hours, the formula is as follows: ; Calculate the average rate of change , as the daily average light change, the formula is as follows: ; When extracting motion features, it is necessary to obtain the activity frequency and intensity of people and vehicles. The number of activities of people and vehicles in each time period is counted. The activity detection data is assumed to be in binary form, 0 indicates no activity, 1 indicates activity, the sampling point 1 of the activity detection data is 1, the sampling point 2 is 0, the sampling point 3 is 1, the sampling point 4 is 0, the sampling point 5 is 1, and the total number of activities is for: ; in, is the number of activities at the i-th sampling point; activity intensity It is measured by calculating the duration of the activity in each time period, and the activity duration is , the total time is T=5, and the formula for activity intensity is as follows: ; When extracting position features, it is necessary to obtain the forklift position and position change speed, where the forklift position is represented as a three-dimensional coordinate , let the forklift position of the i-th sampling point be , then the forklift position It is expressed as: ; Calculate the position change speed in each time period, the position change of the x-axis is , the position of the y-axis changes to , the position change of z axis is , position change speed The calculation formula is as follows: ; but, , , ;in, is the set of position changes on the x-axis, is the set of position changes on the y-axis, is the set of position changes on the z-axis.

[0024] Calculate average speed , as the position change speed, the formula is as follows: ; When extracting temperature features, the current temperature and daily temperature change are required. The current temperature directly uses the temperature value of each sampling point. ; Calculate the rate of change of temperature in each time period , the formula is as follows: ; Calculate the average rate of change , as the daily temperature change, the formula is as follows: ; Combine the features extracted at sampling point i into independent comprehensive feature vectors at different sampling points , expressed as follows: ; The independent comprehensive feature vectors of all different sampling points Combine to form a comprehensive feature vector .

[0025] According to step S200, a multilayer perceptron model is designed, which includes an input layer, two hidden layers and an output layer, wherein the input layer includes n neurons corresponding to the dimension of the comprehensive feature vector, and the first hidden layer includes neurons, and the second hidden layer contains The activation function of the first hidden layer and the second hidden layer is ReLU. The output layer contains m neurons, corresponding to the dimension of the lighting control instructions, and the activation function is Softmax.

[0026] According to step S200, forward propagation is performed, and the input feature vector is a comprehensive feature vector , with a dimension of n, the output of the first hidden layer is ,in, is the weight matrix from the input layer to the first hidden layer, with a dimension of , is the bias vector of the first hidden layer, with dimension ; The output of the second hidden layer is ,in, is the weight matrix from the first hidden layer to the second hidden layer, with a dimension of , is the bias vector of the second hidden layer, with dimension ; The output of the output layer is y, ,in, is the weight matrix from the second hidden layer to the output layer, with a dimension of , is the bias vector of the output layer, with dimension ; Assume the target output is , using mean square error as the loss function for calculation, the formula is as follows: ; in, is the loss function, is the ith predicted output, is the i-th true output.

[0027] According to step S200, the gradient of the loss function to the weights and biases of each layer is calculated by the back propagation algorithm; the gradient calculation formula from the output layer to the second hidden layer is as follows: ; ; in, is the loss function for The gradient of is the loss function for The gradient of is the transpose of the output vector of the second hidden layer; The gradient calculation formula from the second hidden layer to the first hidden layer is as follows: ; in, is the error term of the second hidden layer, yes The transpose of is the derivative of the ReLU activation function of the second hidden layer, represents element-wise product; ; ; in, is the loss function for The gradient of is the loss function for The gradient of is the transpose of the output vector of the first hidden layer; The gradient calculation formula from the first hidden layer to the input layer is as follows: ; in, is the error term of the first hidden layer, yes The transpose of is the derivative of the ReLU activation function of the first hidden layer; ; ; in, is the loss function for The gradient of is the loss function for The gradient of is the transpose of the input vector.

[0028] According to step S200, the weight updating process is as follows: using the gradient descent method, updating the weight and bias of the output layer according to the calculated gradient, updating the weight and bias of the second hidden layer, and updating the weight and bias of the first hidden layer. The step size of the weight update is determined by the learning rate. The learning rate is a pre-set hyperparameter. In this embodiment, the learning rate is set to 0.01; The process of model training is as follows: initialize the weights and biases of all layers to random values; divide all training samples into several batches, and perform forward propagation, loss calculation and back propagation once for each batch; average the gradients of each batch to update the weights and biases; repeat the above steps until the loss function converges or reaches the preset number of iterations; at the end of each iteration cycle, evaluate the performance of the model on the validation set to monitor the training process and prevent overfitting; The trained model performs forward propagation based on the real-time input sensor data to generate corresponding lighting control instructions. The output lighting control instructions are used to adjust the warehouse's lighting system.

[0029] According to step S300, input variables of the lighting system are defined, wherein the input variables of the lighting system include the brightness setting of each lamp, the natural light brightness of each area, and the ambient temperature; an energy consumption model is constructed, including the following factors: the relationship between the power consumption of each lamp and its brightness and the total energy consumption of each area, wherein the total energy consumption of each area is the sum of the energy consumption of all lamps; The brightness setting of each lamp indicates the light output intensity of the lamp, which can be a number between 0 and the maximum brightness value; the natural light brightness of each area indicates the light intensity from natural light sources in a specific area, which is obtained through real-time monitoring by light sensors; the ambient temperature in the warehouse affects the efficiency and life of the lamps, which is obtained through real-time monitoring by temperature sensors.

[0030] In this embodiment, it is assumed that there are four areas (area A, area B, area C, and area D) in the warehouse, each area has two lamps, and the lamp brightness setting represents the light output intensity of the lamp, with a value range of 0 to 100; the initial brightness is set as follows: the brightness of lamp 1 and lamp 2 in area A is 60 and 70 respectively, the brightness of lamp 3 and lamp 4 in area B is 50 and 60 respectively, the brightness of lamp 5 and lamp 6 in area C is 40 and 50 respectively, and the brightness of lamp 7 and lamp 8 in area D is 30 and 40 respectively.

[0031] The natural light brightness is acquired through real-time monitoring by a light sensor. The initial natural light brightness is: 200 lx in area A, 150 lx in area B, 100 lx in area C, and 50 lx in area D.

[0032] The initial ambient temperature was 25°C.

[0033] The power consumption of each lamp is proportional to the brightness. Power consumption (W) = brightness setting / 100 * maximum power. The maximum power of each lamp is 100W. The total energy consumption of each area is the sum of the energy consumption of all lamps in the area. For area A, the initial total energy consumption is calculated as follows: , area B-area D and so on.

[0034] In this embodiment, a multi-objective optimization problem is defined. For the first objective, it is ensured that the illumination intensity of each area is greater than or equal to the minimum illumination requirement, and the minimum illumination requirement is set to 300 lx. For the second objective, the brightness setting of each lamp is optimized to minimize the total energy consumption. The process of using the particle swarm optimization algorithm is as follows: generate a particle swarm, each particle represents a combination of lamp brightness settings; each particle updates the brightness setting and speed based on its own experience and the experience of its neighbors; calculate the fitness of each particle, which is the satisfaction of energy consumption and lighting requirements; update the individual optimal position of each particle and the global optimal position of the group; repeat the above steps until the preset number of iterations is reached; the position of the optimal particle is combined with the lighting control instruction as the optimal lighting control instruction and applied to the warehouse lighting system to adjust the brightness and status of each lamp.

[0035] According to step S400, a minimum threshold of natural light brightness in each area is defined, and a light sensor is used to monitor the natural light brightness in each area in real time. When the natural light brightness in a certain area is lower than the threshold, the system automatically generates a fault alarm and sends an alarm message to the staff, prompting them to check whether the light source is damaged or blocked; The system continuously monitors the brightness of natural light. When it detects that the brightness of natural light in a certain area is insufficient, it automatically increases the brightness of the lamps around the area according to the preset strategy to ensure that the total light intensity meets the requirements. The system monitors the power status of the warehouse in real time. When a power failure is detected, the system automatically switches to the backup power supply. The switching process is seamless to avoid lighting interruptions. A power failure alarm is sent to the staff, prompting them to check the cause of the main power failure.

[0036] According to another embodiment of the present invention, Figure 2 As shown in the system structure diagram of a linkage lighting control system and method based on the Internet of Things, a linkage lighting control system based on the Internet of Things includes the following steps: Data collection and preprocessing module: including: data collection unit and data preprocessing unit; the data collection unit uses sensors to monitor the natural light brightness in various areas of the warehouse, monitor personnel and vehicle activities, track the location of forklifts in real time, and monitor ambient temperature; the data preprocessing unit preprocesses the collected data and transmits it to the local server and cloud server; Feature extraction and model training module: including: feature extraction unit, multi-layer perceptron design unit and model training unit; wherein, the feature extraction unit is responsible for obtaining pre-processed data from the cloud server and extracting features, wherein the features include light intensity features, motion features, position features and temperature features, and combining the extracted features into a comprehensive feature vector; the multi-layer perceptron design unit uses the multi-layer perceptron to train the machine learning model and designs the structure of the multi-layer perceptron; the model training unit is responsible for forward propagation, loss calculation, back propagation and weight update until the model converges and outputs the lighting control instructions; Energy consumption optimization module: including: energy consumption model construction unit, multi-objective optimization problem definition unit and particle swarm optimization algorithm unit; the energy consumption model construction unit is responsible for constructing an energy consumption model to calculate the energy consumption under different lighting settings, the multi-objective optimization problem definition unit is responsible for defining multi-objective optimization problems to balance lighting demand and energy consumption, and the particle swarm optimization algorithm unit solves the multi-objective optimization problem through the particle swarm optimization algorithm, combines the lighting control instructions to obtain the optimal lighting control instructions, and applies them to the lighting system of the warehouse to adjust the brightness and status of each lamp; Fault detection and processing module: including: fault alarm unit, light source adjustment unit and power fault processing unit; among them, the fault alarm unit is responsible for setting the minimum threshold of natural light brightness. When it is detected that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm. When the light source adjustment unit detects that the natural light brightness is insufficient, the system increases the brightness of the surrounding light sources according to the preset strategy. The power fault processing unit is responsible for power fault processing and automatically switching to the backup power supply.

[0037] In this embodiment, the data acquisition and preprocessing module installs light sensors, motion sensors, position sensors and temperature sensors in four areas (A, B, C, and D) of the warehouse. Area A: light intensity 200 lx, personnel activity frequency 10 times / hour, forklift position (5, 10, 0), and ambient temperature 25°C. Area B: light intensity 150 lx, personnel activity frequency 8 times / hour, forklift position (15, 20, 0), and ambient temperature 26°C. Area C: light intensity 100 lx, personnel activity frequency 12 times / hour, forklift position (25, 30, 0), and ambient temperature 24°C. Area D: light intensity 50 lx, personnel activity frequency 15 times / hour, forklift position (35, 40, 0), and ambient temperature 23°C. The collected data is preprocessed, including denoising, smoothing, normalization, etc., and the preprocessed data is transmitted to the local server and the cloud server.

[0038] The feature extraction and model training module obtains preprocessed data from the cloud server and performs feature extraction. The structure of the multilayer perceptron MLP is designed, including the input layer, hidden layer, and output layer. The input layer accepts the comprehensive feature vector, and the hidden layer is set to 3 layers, each containing 10 neurons. The output layer: outputs lighting control instructions. Forward propagation, loss calculation, back propagation, and weight update are performed until the model converges. Output lighting control instructions: including increasing the brightness of the lamps in area A to 70, reducing the brightness of the lamps in area B to 60, etc.

[0039] The energy consumption optimization module builds an energy consumption model. The power consumption of each lamp is proportional to its brightness. Assume that the maximum power of the lamp is 100W. Define a multi-objective optimization problem. Use the particle swarm optimization algorithm for optimization to generate a particle swarm. Each particle represents a combination of lamp brightness settings. Each particle updates the brightness setting and speed based on its own experience and the experience of its neighbors. Calculate the fitness of each particle, consider energy consumption and lighting requirements, update the individual optimal position of each particle and the global optimal position of the group, and repeat the above steps until the preset number of iterations is 100.

[0040] The fault detection and processing module sets the minimum threshold of natural light brightness to 50 lx. When it is detected that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm and notifies the staff. When it is detected that the natural light brightness in a certain area is insufficient, the system increases the brightness of the surrounding lamps according to the preset strategy. For example, when the natural light brightness in area D drops to 40 lx, the system increases the brightness of the lamps in area D to meet the total light intensity requirement. The system monitors the power status of the warehouse in real time.

[0041] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0042] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A linkage lighting control method based on the Internet of Things, characterized in that: The method comprises the following steps: S100, using sensors to monitor the natural light brightness in various areas of the warehouse, monitor personnel and vehicle activities, track the location of forklifts in real time, monitor ambient temperature, pre-process the collected data, and transmit it to local servers and cloud servers; S200, obtaining preprocessed data from a cloud server, performing feature extraction, wherein the features include light intensity features, motion features, position features, and temperature features, and combining the extracted features into a comprehensive feature vector; using a multi-layer perceptron to train a machine learning model, designing a structure of the multi-layer perceptron, performing forward propagation, loss calculation, back propagation, and weight update until the model converges, and outputting a lighting control instruction; S300, build an energy consumption model to calculate the energy consumption under different lighting settings; define a multi-objective optimization problem to balance lighting demand and energy consumption; solve the multi-objective optimization problem through a particle swarm optimization algorithm, combine the lighting control instructions to obtain the optimal lighting control instructions, apply them to the lighting system of the warehouse, and adjust the brightness and status of each lamp; S400. Set the minimum threshold of natural light brightness. When the system detects that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm. When the system detects that the natural light brightness is insufficient, the system increases the brightness of the surrounding light sources according to the preset strategy. It also handles power failures and automatically switches to backup power.

2. The method for controlling linkage lighting based on the Internet of Things according to claim 1, characterized in that: According to step S100, light sensors are installed in various areas of the warehouse to monitor the brightness of natural light, motion sensors are installed in main passages and work areas to monitor the activities of personnel and forklifts, position sensors are installed in forklifts and other transportation equipment to track positions in real time, temperature sensors are installed in various areas of the warehouse to monitor ambient temperature, and the data sampling frequency of each sensor is set; Use median filtering to remove outliers and noise in sensor data, perform outlier detection on data that exceeds the normal range, use interpolation to process missing data points, and normalize or standardize sensor data to unify the dimensions of the data; Through the IoT gateway, the pre-processed sensor data is transferred to the local server for storage; The pre-processed data is synchronously transmitted to the cloud server; during the data transmission process, SSL / TLS is used to ensure data security, encrypt the stored data, and set access control policies.

3. The method for controlling linkage lighting based on the Internet of Things according to claim 1, characterized in that: According to step S200, when extracting the light intensity feature, it is necessary to obtain the current light intensity and the daily average light change, wherein the current light intensity directly uses the light intensity value of each sampling point , where n is the total number of sampling points, They are the light intensity values ​​from sampling point 1 to sampling point n respectively; to obtain the daily average light change, it is necessary to calculate the rate of change of light intensity in each time period , let each time period be , the formula is as follows: ； Calculate the average rate of change , as the daily average light change, the formula is as follows: ； When extracting motion features, it is necessary to obtain the activity frequency and intensity of people and vehicles. The number of activities of people and vehicles in each time period is counted. The activity detection data is in binary form, 0 means no activity, 1 means activity, and the total number of activities is for: ； in, is the number of activities at the i-th sampling point; activity intensity It is measured by calculating the duration of the activity in each time period, and the activity duration is , the total time is T, and the formula for activity intensity is as follows: ； When extracting position features, it is necessary to obtain the forklift position and position change speed, where the forklift position is represented as a three-dimensional coordinate , let the forklift position of the i-th sampling point be , then the forklift position It is expressed as: ; Calculate the position change speed in each time period, the position change of the x-axis is , the position of the y-axis changes to , the position change of z axis is , position change speed The calculation formula is as follows: ； Calculate average speed , as the position change speed, the formula is as follows: ； When extracting temperature features, the current temperature and daily temperature change are required. The current temperature directly uses the temperature value of each sampling point. ; Calculate the rate of change of temperature in each time period , the formula is as follows: ； Calculate the average rate of change , as the daily temperature change, the formula is as follows: ； Combine the features extracted at sampling point i into independent comprehensive feature vectors at different sampling points , expressed as follows: ； The independent comprehensive feature vectors of all different sampling points Combine to form a comprehensive feature vector .

4. The method for controlling linkage lighting based on the Internet of Things according to claim 3 is characterized in that: According to step S200, a multilayer perceptron model is designed, which includes an input layer, two hidden layers and an output layer, wherein the input layer includes n neurons corresponding to the dimension of the comprehensive feature vector, and the first hidden layer includes neurons, and the second hidden layer contains The activation function of the first hidden layer and the second hidden layer is ReLU. The output layer contains m neurons, corresponding to the dimension of the lighting control instructions, and the activation function is Softmax.

5. The method for controlling linkage lighting based on the Internet of Things according to claim 4 is characterized in that: According to step S200, the process of forward propagation is as follows: the input feature vector enters the input layer, the output of the input layer is passed to the first hidden layer, in the first hidden layer, each neuron receives the output of the input layer, performs weighted summation and adds a bias, and applies the activation function ReLU to obtain the output of the first hidden layer; The output of the first hidden layer is passed to the second hidden layer. In the second hidden layer, each neuron receives the output of the first hidden layer, performs weighted summation and adds a bias, and applies an activation function to obtain the output of the second hidden layer. The output of the second hidden layer is passed to the output layer. In the output layer, each neuron receives the output of the second hidden layer, performs weighted summation and adds a bias to obtain the final model prediction output. Calculate the gap between the model's predicted output and the true output, and use the mean squared error as the loss function to evaluate the gap and get the loss value.

6. The method for controlling linkage lighting based on the Internet of Things according to claim 5, characterized in that: According to step S200, the process of back propagation is as follows: calculating the error term of the output layer, the difference between the predicted output and the true output; calculating the gradient of the weight and bias from the output layer to the second hidden layer; calculating the error term of the second hidden layer, which is transmitted back from the error term of the output layer through the weight of the output layer and combined with the derivative of the activation function of the second hidden layer; Calculate the gradient of the weights and biases of the second hidden layer to the first hidden layer; calculate the error term of the first hidden layer, which is the error term of the second hidden layer propagated back through the weights of the second hidden layer and combined with the derivative of the activation function of the first hidden layer; Compute the gradients of the weights and biases of the first hidden layer to the input layer.

7. The method for controlling linkage lighting based on the Internet of Things according to claim 6, characterized in that: According to step S200, the weight updating process is as follows: using the gradient descent method, updating the weight and bias of the output layer according to the calculated gradient, updating the weight and bias of the second hidden layer, and updating the weight and bias of the first hidden layer, the step size of the weight update is determined by the learning rate, and the learning rate is a pre-set hyperparameter; The process of model training is as follows: initialize the weights and biases of all layers to random values; divide all training samples into several batches, and perform forward propagation, loss calculation, and backpropagation once for each batch; average the gradients of each batch to update the weights and biases; repeat the above steps until the loss function converges or reaches the preset number of iterations; At the end of each iteration, evaluate the performance of the model on the validation set to monitor the training process and prevent overfitting; The trained model performs forward propagation based on the real-time input sensor data to generate corresponding lighting control instructions. The output lighting control instructions are used to adjust the warehouse's lighting system.

8. The method for controlling linkage lighting based on the Internet of Things according to claim 1, characterized in that: According to step S300, the input variables of the lighting system are defined, wherein the input variables of the lighting system include the brightness setting of each lamp, the natural light brightness of each area, and the ambient temperature; and an energy consumption model is constructed, including the following factors: The power consumption of each luminaire in relation to its brightness and the total energy consumption of each zone, where the total energy consumption of each zone is the sum of the energy consumption of all luminaires; Define a multi-objective optimization problem. For the first objective, ensure that the light intensity of each area is greater than or equal to the minimum light requirement. For the second objective, optimize the brightness setting of each lamp to minimize the total energy consumption. The process of using the particle swarm optimization algorithm is as follows: generate a particle swarm, each particle represents a combination of lamp brightness settings; each particle updates the brightness setting and speed based on its own experience and the experience of its neighbors; calculate the fitness of each particle, which is the satisfaction of energy consumption and lighting requirements; update the individual optimal position of each particle and the global optimal position of the group; repeat the above steps until the preset number of iterations is reached; the position of the optimal particle is combined with the lighting control instruction as the optimal lighting control instruction and applied to the warehouse lighting system to adjust the brightness and status of each lamp.

9. The method for controlling linkage lighting based on the Internet of Things according to claim 1, characterized in that: According to step S400, a minimum threshold of natural light brightness in each area is defined, and a light sensor is used to monitor the natural light brightness in each area in real time. When the natural light brightness in a certain area is lower than the threshold, the system automatically generates a fault alarm and sends an alarm message to the staff, prompting them to check whether the light source is damaged or blocked; The system continuously monitors the brightness of natural light. When it detects that the brightness of natural light in a certain area is insufficient, it automatically increases the brightness of the lamps around the area according to the preset strategy to ensure that the total light intensity meets the requirements. The system monitors the power status of the warehouse in real time. When a power failure is detected, the system automatically switches to the backup power supply. The switching process is seamless and avoids lighting interruptions. Send a power failure alarm to the staff, prompting them to check the cause of the main power failure.

10. A linkage lighting control system based on the Internet of Things, using a linkage lighting control method based on the Internet of Things according to any one of claims 1 to 9, characterized in that: The following steps are involved: Data collection and preprocessing module: including: data collection unit and data preprocessing unit; the data collection unit uses sensors to monitor the natural light brightness in various areas of the warehouse, monitor personnel and vehicle activities, track the location of forklifts in real time, and monitor ambient temperature; the data preprocessing unit preprocesses the collected data and transmits it to the local server and cloud server; Feature extraction and model training module: including: feature extraction unit, multi-layer perceptron design unit and model training unit; wherein, the feature extraction unit is responsible for obtaining pre-processed data from the cloud server and extracting features, wherein the features include light intensity features, motion features, position features and temperature features, and combining the extracted features into a comprehensive feature vector; the multi-layer perceptron design unit uses the multi-layer perceptron to train the machine learning model and designs the structure of the multi-layer perceptron; the model training unit is responsible for forward propagation, loss calculation, back propagation and weight update until the model converges and outputs the lighting control instructions; Energy consumption optimization module: including: energy consumption model construction unit, multi-objective optimization problem definition unit and particle swarm optimization algorithm unit; the energy consumption model construction unit is responsible for constructing an energy consumption model to calculate the energy consumption under different lighting settings, the multi-objective optimization problem definition unit is responsible for defining multi-objective optimization problems to balance lighting demand and energy consumption, and the particle swarm optimization algorithm unit solves the multi-objective optimization problem through the particle swarm optimization algorithm, combines the lighting control instructions to obtain the optimal lighting control instructions, and applies them to the lighting system of the warehouse to adjust the brightness and status of each lamp; Fault detection and processing module: including: fault alarm unit, light source adjustment unit and power fault processing unit; among them, the fault alarm unit is responsible for setting the minimum threshold of natural light brightness. When it is detected that the light intensity in a certain area is lower than the threshold, the system automatically generates a fault alarm. When the light source adjustment unit detects that the natural light brightness is insufficient, the system increases the brightness of the surrounding light sources according to the preset strategy. The power fault processing unit is responsible for power fault processing and automatically switching to the backup power supply.

Images