Intelligent control and energy saving evaluation of highway tunnel lighting: Based on three-dimensional simulation and long short-term memory optimization algorithm

Traditional tunnel lighting luminance is calculated based on the designed hourly traffic volume and designed speed. This approach first determined the entrance section reduction factor and middle section's luminance through the look-up table method, then calculated each section's luminance in proportion. Since the designed value is often higher than the actual traffic flow value, and the calculated brightness value is so high which might lead to tunnel energy consumption. What's more, the designed value in this method is the piecewise discrete value, which does not meet the requirement of continuous dimming for LED lighting. In this case, under the premise of satisfying operation safety, how to carry out continuous intelligent energy saving control research and energy saving evaluation based on actual traffic value is an urgent problem to be solved in tunnel lighting. This paper first proposed a luminance calculation model for each lighting section which can realize continuous dimming as needed. Using the least squares method fitting the reduction coefficient and the luminance of the middle section to solve the luminance of the entrance and middle sections under the condition of any speed and traffic flow, and realizing the continuous optimization of the luminance calculation. Second, in terms of the shortcomings of the traditional graded dimmer control, this paper offered a highway tunnel lighting intelligent control algorithm with traffic flow, speed and luminance out of the tunnel as the input matrix, which meets the requirement of LSTM (Long Short-Term Memory) neural networks, tunnel lighting characteristics and tunnel lighting section. Considering the objective function of optimization problem in LSTM, three gradient descent methods were introduced to optimize the model respectively. Third, this paper adopted equal proportion to arrange lamps and light distribution, and developed a tunnel lighting simulation environment based on DIALux to meet the energy conservation assessment requirements in operating tunnels. Roadway luminance and its total uniformity are selected as energy saving evaluation indexes. Finally, we established a three-dimensional tunnel model with the real environment of Jinding Lake 2# tunnel, then conducted simulations and experimental verifications of the intelligent control algorithm under sunny and cloudy weather conditions, and verified the performance of the intelligent lighting control algorithm under the three optimizers. The result of energy conservation evaluations shows that compared with traditional lighting design, the tunnel lighting energy conservation algorithm can save 23.61% of energy in sunny days and 31.40% in cloudy days.