Using the Internet of Things, artificial intelligence, big data and other technologies, designed a set of industrial plant intelligent lighting system.
The system adopts the management platform, intelligent lamp controller, intelligent LED device lamp three-layer architecture mode, the comprehensive use of video acquisition, light sensing, deep learning network and other technologies, to achieve the industrial plant lighting field of constant light intelligent non-level dimming and comprehensive management.
In order to meet the enterprise fine management plant lighting at the same time, but also to obtain significant energy-saving effect.
Industrial plant in the manufacturing process plays an important role, has been, industrial plant lighting lamps are mostly high-pressure sodium lamps, but its short life, high energy consumption, non-dimming and other shortcomings are increasingly prominent. With the vigorous promotion of intelligent manufacturing, energy conservation and emission reduction, industrial LED lighting with its unique low power consumption, long life and other characteristics of rapid entry into the industry’s vision, and with the growing maturity of the LED industry, its reliability and cost-effective gradually reflected, now, more and more industrial and mining enterprises lighting will be preferred LED products.
Intelligent plant can not be separated from intelligent lighting, LED with the advantages of non-destructive level dimming so that the plant 24 hours a day to maintain constant illumination lighting provides technical feasibility. Constant illumination means that the lighting values at the same height are relatively consistent at any time in the plant, and constant illumination provides the best ambient lighting in the plant, and each LED luminaire maintains only the minimum brightness under constant light conditions, thus minimizing the energy consumption of lighting in the plant. From the software and hardware level design and construction of a complete set of industrial plant intelligent lighting system, and then realize the industrial plant lighting field of constant light intelligent stepless dimming and comprehensive management.
Industrial plant intelligent LED Indurstrial Lighting system consists of a lighting management platform, a number of intelligent lamp controllers, a number of single-light controllers, LED industrial and mining lamps, a number of high-definition light environment cameras, a number of light sensitivity sensors, switches and other equipment. The intelligent lighting system architecture for industrial plants is shown in Figure 1. The entire system is installed in a large plant, wherein the communication network consisting of the system is a local area network.
The lighting management platform is a comprehensive management platform for the display and control of the lighting system, which is installed in the server and connected to the network switch by wire. Management computers and mobile phone APPS can be connected to the lighting management platform via network switches and LAN Wi-Fi for viewing and parameter settings on various devices in the lighting system.
Fig.1 Structure of intelligent lighting system for industrial plant
Several intelligent luminaire controllers are connected to the network switch through optical fiber to receive the instructions of the platform and the upload of the lamp data, and the fiber-optic transceiver needs to be set up at both ends of the fiber; Light environment camera and illumination sensor through the intelligent luminaire controller to upload data information to the server in real time, to monitor and analyze the light environment parameters of each area, the addition of light sensitivity sensor is used for camera acquisition and analysis of the optical environment parameters calibration; A smart LED industrial lamp consists of an LED industrial and mine lamp and a single lamp controller, which is used to detect the working status of LED industrial and mine lamps, to count energy consumption in real time, to receive instructions from intelligent lamp controllers and to control the brightness of LED industrial and mine lamps.
An area only needs to be equipped with a smart luminaire controller, a light environment camera, a light sensitivity sensor, several smart LED industrial and mine lamps, a smart lamp controller with all the intelligent LED industrial and mine lights in the area, intelligent luminaire controller can adjust any smart LED industrial and mine lamp any brightness.
1.1、Lighting management platform
Lighting management platform is a comprehensive management platform of intelligent lighting system of the entire industrial plant, which can view, manage and control the status, brightness, energy consumption and other parameters of all intelligent LED industrial and mining lamps. The lighting management platform is shown in Figure 2. Designed and developed using a “browser/server” (B/S, Browser/Server) architecture, the platform features high reliability, security, stability, ease of use, and more. Functions include: user role rights management, real-time viewing of lamp status, remote adjustment of lamp brightness, lamp levelless dimming, intelligent control, partition group management, time cycle planning, real-time push of lamp failure, big data display energy consumption information.
Fig.2 Lighting management platform
1.2、Intelligent control model
Light environment camera real-time capture responsible area of environmental photos and upload to the lighting management platform, the platform uses deep learning algorithm to calculate the lighting degree of each place in the area, according to the constant light preset value, the platform after calculation will issue different instructions to control the intelligent LED industrial and mine lamp adjusted to different brightness, in order to achieve the whole area of constant illumination lighting effect. The most critical part of this is extracting light from ambient photos, so the platform incorporates deep learning algorithms in artificial intelligence and uses convolutional neural networks (CNN) to efficiently and quickly extract light across the environment.
Convolutional neural network has strong feature expression ability, and has good robustness and generalization, and can learn the light features and deep features such as color, texture, shape and glossy surface in the image by itself, and has high expressiveness for the extraction of ambient light. As shown in Figure 3, the image on the far left is the input layer, followed by the convolution layer, followed by the pooling layer, which can be repeatedly multigrouped as a hidden layer in the middle of the model, with two groups appearing, followed by the Full Connected Layer, followed by the output layer Predictions Layer)。
Convolution layer is responsible for image illumination extraction features, pooled layer is responsible for feature selection, and full connection layer is responsible for the classification of illumination values. The extraction process of the illumination of each area is as follows: In the convolution layer, we need to construct a suitable set of environmental feature filters in advance, the filter and the input image convolution in the convolution layer to obtain a set of feature vector sets, in the pool layer for light points Feature selection, culling the parameters of equipment objects in the environment, selecting more suitable ambient illumination feature points, and then, through reverse propagation, infinitely fit the feature vector set that can represent the illumination of the real environment, and then classify these feature points through the full connection layer. The light intensity values of different regions are divided according to the brightness level.
Fig.3 Convolution neural networks model
2 、Experimental testing
In the experimental link, we apply this intelligent lighting system to a large industrial plant. The overall layout of the plant is rectangular, 420 long, 120 wide, 20 high, building area of about 50,000, plant for large logistics storage and transfer, 24 hours a day operation, the top of the plant without natural light panels, four walls of the wall spaced installation windows, the overall proportion of 30%, for natural lighting during the day.
According to the actual layout of the plant, and in accordance with the national building lighting design standards, the plant is divided into 14 partitions, as shown in Figure 4, each partition length and width are 60, in each partition, placed a smart lamp controller, 1 light environment camera, 1 light Degree sensor, 49 intelligent industrial and mine lamps, 7×7 points evenly arranged, plant partition as follows 4 partition 8, a single lamp power 200, by the simulation of the previous experiments, at night in the lamp fully open and full power operating mode, The average luminance value of 0.75 standard horizontal surface is 300, which is better than the design standard 200, and the margin design takes into account special lighting needs and later lamp light decay.
Therefore, the entire plant has a total of: 14 intelligent lamp controller, 14 light environment cameras, 14 light sensitivity sensors, 686 intelligent industrial and mining lamps. Because of the plant storage logistics, direct sunlight and temperature have special requirements, so the plant does not have a roof natural light board, only the four walls of the plant have windows to light natural light, so even in the daytime, the plant interior is also required to open part of the lighting for the plant interior lighting.
Fig.4 Plant zoning
Take Section 8 in Figure 4 as an example, the left and bottom sides of Zone 8 are windowed, when all the lamps in the plant are closed, the light level of each point in the score area, natural light under partition 8 in the area of light distribution as shown in Figure 5. The horizontal coordinates represent the horizontal distance of the partition, and the ordinates represent the vertical distance of the partition. As can be seen from the figure, the light value is higher in the area near the window side, lower in the area away from the window side, and lower in the farther away from the window.
Run intelligent lighting system, set the constant light target value of 200 lx, after the system operates autonomously, after continuous analysis, calculation and control, the lighting of all areas of the plant to maintain relative consistency, to achieve the effect of constant light. Also take Partition 8 in Figure 4 as an example, after the system has been running for some time, the light distribution of each area in Partition 8 in intelligent dimming mode is shown in Figure 6. The horizontal coordinates represent the horizontal distance of the partition, and the ordinates represent the vertical distance of the partition.
As can be seen from the figure, after the computer intelligently controls the luminous brightness of each lamp, the areas under the partition, whether close to or away from the window side, the illumination of each area remains approximately the same within the same horizontal surface. It should be noted that the value fluctuates, because in the actual light-emitting process of lamps, by the luminous angle and space gas transmission and other factors, the general error is within the upper and lower 20%, are in line with the design standards, belong to the category of constant light.
Fig.5 Illuminance distribution of each area in Zone 8 under natural light illumination
Any choice of weather conditions, similar natural light, plant operating time is equivalent (all 24 hours a day), work busy degree is similar to two days, using two lighting control methods to do energy consumption comparison experiments, one control method for no intelligent lighting scheme, that is, the traditional manual control method, lighting switch using physical open way, lamps can not be intelligent dimming, during the day can only be artificially based on the field operation to open a small number of lamps, open most lamps at night; Another way of control is to use the intelligent lighting system of the industrial plant to automatically control the brightness of each lamp according to the characteristics of the on-site environment in the plant. The distribution of power consumption for each time period under different lighting control schemes is shown in Figure 7.
As can be seen from the figure, in each time of day, no intelligent lighting scheme (i.e., traditional manual control methods), day and night power consumption are their own fixed values, because the number of lights on during the day is certain, and the number is smaller, the power consumption is relatively low, the number of lights turned on at night is more, the power consumption is relatively high.
Intelligent lighting scheme, each time period of electricity consumption is dynamic changes, and show a relatively smooth trend, because the intensity of natural light during the day will change slowly over time, the system will be based on real-time intelligent adjustment of the brightness of each lamp, daytime lamp brightness will be adjusted relatively low, sooner or later relatively high, night will be higher, in order to achieve a factory all-weather relative to constant illumination lighting effect. At the same time precise adjustment will precisely save every piece of electricity, compared with the traditional manual control method, is a fine management method, the overall effective energy saving in more than 20%.
Fig.6 Illuminance distribution of each area in Zone 8 under intelligent dimming mode
Fig.7 Distribution of electric energy consumption of different lighting control schemes
3、The advantages of intelligent industrial lighting systems
The application of intelligent lighting system in industrial plant will significantly improve the intelligent, precise, refined, humanized, efficient and scientific management of lighting equipment in industrial plant.And for large-scale plant enterprises to bring considerable economic benefits, management benefits and social benefits.
The brightness of each lamp will be calculated by the computer based on a comprehensive analysis of the environmental parameters in the plant. In order to improve the accuracy of the calculation results, the comprehensive analysis of environmental parameters uses artificial intelligence deep learning algorithm to analyze the environmental video stream in real time quickly and efficiently, and calculate the most suitable illumination of each area.
After the platform server calculates the most needed light level in each area, the brightness of each lamp is adjusted precisely and smoothly by intelligent luminaire controller and single lamp controller in a senseless manner, and the whole control process is fully automated, without human intervention, and further enhances the effectiveness of scientific management.
3.3、Fine energy saving
For each lighting system in the plant will calculate a different brightness and control, such as the day near the natural light area of the light will be a little darker, away from the natural light area of the light will be a little brighter, at night more goods area lights will be a little light, less cargo area light will be a little darker, so that the plant in the whole area of constant light at the same time to maximize the cost of saving each piece of electricity, reduce the cost of enterprise operation.
Human body non-sensational fine lamp control process, so that the employees in it will not feel visual fatigue due to the lighting and dark, further enhance the employee’s environmental visual comfort, promote the efficiency of staff production and operation.
The system monitors the operating status of each lamp in real time, reports and actively analyzes the cause of the fault for fault, zero delay, effectively reducing the difficulty of troubleshooting the maintenance personnel, improving the efficiency of the repair and reducing the maintenance cost.
Reducing the over-participation of human factors is an effective means to improve scientific management, and the system can operate independently and steadily after the parameter setting is completed. At the same time, real-time status display, energy consumption big data analysis, equipment management, rights management and other functions for scientific management to provide the necessary information support.
Industrial plant intelligent lighting system by the lighting management platform, intelligent lamp controller, single-channel controller, LED industrial and mining lamps, high-definition cameras, lighting sensors and other equipment. On the basis of making full use of natural light, artificial intelligence algorithm is used to analyze the light environment parameters in the plant in real time, adaptively intelligently adjust the brightness of the lamps, and keep the lighting state of the plant at all times.
Scientific lighting management methods bring a higher level of humanized environmental visual experience, effectively promote the efficiency of production operations; Precise control methods and maximize energy savings, for enterprises to bring greater cost-effectiveness and economic benefits.