Street lamp group autonomous scheduling method and device based on improved game theory, and medium

Abstract

The invention discloses a street lamp group autonomous scheduling method and device based on an improved game theory and a medium, belongs to the technical field of intelligent street lamps, and is used for solving the technical problem that an existing street lamp scheduling method lacks comprehensive analysis capability of various environmental data and lacks dynamic adjustment and optimization capability. The method comprises the steps of collecting brightness data, traffic flow data and people flow density data of any street lamp node in an intelligent street lamp group through an Internet of Things technology, and constructing an environment data set; based on the environment data set, constructing an improved game theory model; based on the improved game theory model, determining an optimal operation strategy of any street lamp node; and sending the optimal operation strategy to a corresponding street lamp node, and indicating the street lamp node to adjust the illumination intensity and / or the on-off state based on the optimal operation strategy. According to the method, the energy efficiency and the response capability of the intelligent street lamp group are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart street lamps, and in particular to a method, device and medium for autonomous scheduling of a street lamp group based on improved game theory. Background Art

[0002] Smart city is the direction of future urban development, and one of its core is the smart street light system. Traditional street light systems usually rely on fixed schedules or simple sensor feedback for control, which has many shortcomings. Street light systems controlled by fixed schedules cannot be flexibly adjusted according to changes in the actual environment, resulting in high energy consumption and serious waste of resources. In actual applications, in many cases, there are few vehicles and pedestrians on the road, but high brightness lighting is still maintained, resulting in power waste. Although the control system with simple sensor feedback has improved, it still lacks dynamic adjustment and adaptive capabilities, and it is difficult to cope with the complex and changing urban environment.

[0003] In the existing technology, some smart street light systems have begun to use sensor-based data feedback for lighting scheduling. These systems adjust the switch status and lighting intensity of street lights by detecting the brightness, traffic flow and crowd density data on the road. However, these systems still have obvious defects, which are mainly manifested in the following aspects:

[0004] First, most existing street light dispatching systems rely on single sensor data for simple decision-making and cannot effectively combine multiple environmental variables for comprehensive analysis. For example, only adjusting the lighting intensity through the brightness sensor, while ignoring the changes in traffic flow and pedestrian density, this single data source decision-making method can easily lead to unsatisfactory lighting effects or excessive energy consumption.

[0005] Secondly, existing systems lack flexibility and intelligence in responding to environmental changes. Traditional control algorithms are usually based on preset rules or simple feedback mechanisms and cannot be dynamically adjusted and optimized according to real-time environmental data. For example, when there are sudden changes in traffic flow or pedestrian density on the road, it is difficult for the system to respond in time, resulting in poor lighting effects or waste of resources.

[0006] In summary, the defects of the existing technology are mainly concentrated in the following aspects: lack of comprehensive analysis capability of various environmental data, lack of dynamic adjustment and optimization capabilities. Summary of the invention

[0007] The present invention provides a street lamp group autonomous scheduling method, device and medium based on improved game theory, which are used to solve the technical problems that the existing street lamp scheduling method lacks the comprehensive analysis ability of multiple environmental data and lacks the dynamic adjustment and optimization ability.

[0008] In a first aspect, the present invention provides a street lamp group autonomous scheduling method based on improved game theory, the method comprising:

[0009] Through the Internet of Things technology, the brightness data, traffic flow data and crowd density data of any street light node in the smart street light group are collected to build an environmental data set;

[0010] Based on the environmental data set, construct an improved game theory model;

[0011] Based on the improved game theory model, determining the optimal operation strategy of any street light node;

[0012] The optimal operation strategy is sent to the corresponding street light node, instructing the street light node to adjust the lighting intensity and / or switch state based on the optimal operation strategy.

[0013] In a feasible implementation manner, based on the improved game theory model, determining the optimal operation strategy of any street light node includes:

[0014] Based on the improved game theory model, an initial strategy set is allocated to any street lamp node, wherein any strategy in the initial strategy set includes different lighting intensities and switch states;

[0015] Dynamically adjust the strategy weight of any strategy based on the performance of any street lamp node under different environmental data sets;

[0016] Based on the strategy weight of any one of the strategies, the optimal operation strategy of any one of the street lamp nodes is determined.

[0017] In a feasible implementation manner, based on the performance of any street lamp node under different environmental data sets, dynamically adjusting the policy weight of any policy includes:

[0018] At each time step t, calculating the instantaneous return of any of the strategies according to the brightness data, traffic flow data and crowd density data in the environmental data set;

[0019] According to the formula Adjust the policy weight of any of the above policies, where w ij (t) is the strategy s ij The policy weight at time step t, R ij (t) is strategy s ij Immediate returns, For strategies ij The expected return at time t, η1 is the learning rate;

[0020] Based on the strategy weight of any strategy, determining the best operation strategy of any street lamp node includes: making w ij The strategy with the largest (t+1) is determined as the optimal operation strategy for any street lamp node.

[0021] In a feasible implementation manner, determining the optimal operation strategy of any street lamp node based on the improved game theory model further includes:

[0022] Construct an adaptive learning model based on the combination of deep Q network and long short-term memory network;

[0023] Initializing the parameters of the deep Q network, defining a state space, an action space, and a reward function, wherein the state space includes brightness data, traffic flow data, and crowd density data;

[0024] Using the deep Q network, using the environment data set to perform strategy learning, and calculating the Q value of any action in the action space;

[0025] At each time step t, the environment data set is input into the long short-term memory network, the temporal features of the environment data set are extracted, and the implicit state of the environment data set is output:

[0026] Determining a comprehensive evaluation value of any one of the actions based on the Q value and the implicit state;

[0027] Based on the comprehensive evaluation value, determining the optimal action in the action space;

[0028] Instruct any street lamp node to adjust its own strategy selection according to the optimal action and continuously optimize the strategy. The learning process of the optimization strategy adopts experience playback and target network technology.

[0029] In a feasible implementation manner, based on the improved game theory model, determining the optimal operation strategy of any street lamp node further includes:

[0030] Based on the dynamic alliance formation theory, the street light nodes in the smart street light group are instructed to form a temporary alliance according to the real-time environmental data and their own status, and the cooperation benefit function U of the two street light nodes in the temporary alliance is calculated. ij (t) = β·E i (t)+γ·E j (t)-δ·C ij (t), where β and γ are the energy efficiency weights of street light node i and street light node j, E i (t) and E j (t) is the energy efficiency of street light node i and street light node j, C ij (t) is the collaboration cost between streetlight node i and streetlight node j, δ is the collaboration cost weight;

[0031] In high-traffic areas, the street light nodes in the smart street light group are instructed to form an alliance C(t) to jointly adjust the lighting intensity to achieve the best lighting effect and energy-saving goals. The strategy optimization function within each alliance is Among them, η C is the learning rate of the alliance C(t), is the coalition policy gradient.

[0032] In a feasible implementation manner, based on the improved game theory model, determining the optimal operation strategy of any street lamp node further includes:

[0033] An anomaly detection module is embedded in the environmental data set, and the anomaly detection module is used to monitor the status of any street lamp node in real time;

[0034] Calculate the abnormality detection function of any street light node, and when the abnormality detection function is greater than a preset abnormality threshold, determine that any street light node is in an abnormal state, and the abnormality detection function Where M is the number of monitoring parameters, is the current value of the kth parameter, is the historical mean of the kth parameter;

[0035] When any of the street light nodes is detected to be in an abnormal state, the abnormality detection module is used to automatically adjust the strategy of the street light node;

[0036] The anomaly detection module is also used to send alarm information to the central control center, and the alarm information includes the number of the street light node, the anomaly type and the current environment data set.

[0037] In a feasible implementation manner, instructing the street lamp node to adjust the lighting intensity and / or switch state based on the optimal operation strategy includes:

[0038] Based on the formula Instruct any street lamp node to adjust the current lighting intensity to L based on the optimal operation strategy i (t), where L i (t) represents the actual lighting intensity of the i-th street light node at time t, L max Represents the maximum illumination intensity, α i Represents the adjustment coefficient used to control the response degree of the current street light node to the optimal lighting intensity. represents the optimal lighting intensity calculated by the i-th street light node at time t, β i represents the adjustment coefficient used to control the degree of dependence of the current street light node on the lighting intensity of adjacent nodes, N(i) represents the set of nodes adjacent to the i-th street light node, and w ijrepresents the weight coefficient between street light node i and its adjacent street light node j, L j (t-1) is the lighting intensity of the adjacent street lamp node j at time t-1; and / or,

[0039] Based on the formula Instruct any street lamp node to adjust the current switch state to 0 based on the optimal operation strategy i (t), where O i (t) represents the switch state of the i-th street light node at time t, 1 represents the on state, 0 represents the off state, γ i is the adjustment coefficient used to control the response degree of the current street light node to the optimal switching state. is the optimal switch state calculated by the i-th street light node at time t, δ i is the adjustment coefficient used to control the degree of dependence of the current street light node on the switch state of the adjacent node, v ij is the switch state weight coefficient between street light node i and its adjacent street light node j, O j (t-1) is the switch state of adjacent node j at time t-1.

[0040] In a feasible implementation, based on the formula L′ i (t) = f s (L i (t)), obtain the feedback lighting intensity L′ of any street light node fed back by the lighting intensity sensor i (t), where f s is a sensor feedback function of the illumination intensity sensor;

[0041] Calculate the feedback illumination intensity L′ i (t) and the expected value of the lighting intensity The lighting intensity deviation between i (t);

[0042] Based on the formula Instruct any street lamp node to adjust the lighting intensity, where k L is the adjustment coefficient of lighting intensity and switch state, ξ is the influence coefficient of adjacent nodes on lighting state adjustment; and / or,

[0043] Based on the formula O′ i (t) = g s (O i (t)), obtain the feedback switch state O′ of any street lamp node fed back by the switch state sensor i (t), where g s is a sensor feedback function of the switch state sensor;

[0044] Calculate the feedback switch state O′ i (t) and expected value of switch state The switching state deviation between i (t);

[0045] Based on the formula Instruct any street light node to adjust the switch state, where k O is the adjustment coefficient of the switch state, and ζ is the influence coefficient of the adjacent nodes on the switch state adjustment.

[0046] In a second aspect, the present invention provides an autonomous scheduling device for a group of street lights based on improved game theory, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute a method for autonomous scheduling of a group of street lights based on improved game theory as described in any of the above-mentioned embodiments.

[0047] In a third aspect, the present invention provides a non-volatile computer storage medium, which is a non-volatile computer-readable storage medium. The non-volatile computer-readable storage medium stores at least one program, each of which includes instructions. When the instructions are executed by a terminal, the terminal executes a method for autonomous scheduling of street light groups based on improved game theory as described in any of the above-mentioned embodiments.

[0048] The present invention provides a street lamp group autonomous scheduling method, device and medium based on improved game theory, which has the following beneficial technical effects compared with the prior art:

[0049] (1) The present invention can collect and comprehensively analyze the environmental data set of any street lamp node in the smart street lamp group in real time by introducing the Internet of Things technology, and apply the improved game theory model to determine the optimal operation strategy, so that the system can make the optimal lighting intensity and / or switch state adjustment according to the real-time environmental data (environmental data set), thereby significantly improving the energy efficiency and responsiveness of the smart street lamp group.

[0050] (2) The present invention allocates an initial strategy set to each street light node, and through an improved game theory model, dynamically adjusts the weight of each strategy according to the changes in historical data and real-time data. By designing a collaborative competition mechanism, adjacent street light nodes cooperate in competition to ensure the overall lighting effect while maximizing the energy saving effect. Based on the theory of dynamic alliance formation, street light nodes can form temporary alliances based on real-time environmental data and their own status, share information and resources, and jointly formulate the best strategy. Through this collaborative competition mechanism, the system can achieve global optimization and improve the overall lighting effect and energy utilization efficiency.

[0051] (3) The present invention adopts an adaptive learning model that combines a deep Q network and a long short-term memory network, and combines reinforcement learning and deep learning techniques to continuously optimize the strategy selection of each street light node. By constructing an adaptive learning model, the system can continuously self-learn and adjust according to environmental changes to ensure that each street light node can make the best decision in different environments. The deep Q network is used to calculate the Q value of each action, and the long short-term memory network is used to extract the time series characteristics of the data. The two are combined to calculate the comprehensive evaluation value of each action, thereby selecting the optimal strategy. Through experience replay and target network technology, the learning stability and efficiency are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0053] Figure 1 A flowchart of a street lamp group autonomous scheduling method based on improved game theory provided by the present invention;

[0054] Figure 2 A flow chart of another street lamp group autonomous scheduling method based on improved game theory provided by the present invention;

[0055] Figure 3 A schematic structural diagram of a street lamp group autonomous scheduling method device based on improved game theory provided by the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. 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 making creative work are within the scope of protection of the present invention.

[0057] The technical solution proposed by the present invention is described in detail below with reference to the accompanying drawings.

[0058] Figure 1 The flowchart of the street lamp group autonomous scheduling method based on improved game theory provided by the present invention. Figure 1 As shown, the method includes the following execution steps:

[0059] S10. Through the Internet of Things technology, the brightness data, traffic flow data and pedestrian density data of any street lamp node in the smart street lamp group are collected to build an environmental data set.

[0060] In some embodiments, S10 collects brightness data, traffic flow data, and crowd density data of any street light node in the smart street light group through Internet of Things technology, and constructing an environmental data set includes the following steps.

[0061] S11, collect brightness data L in real time through the light sensor installed on each smart street lamp i (t), where L i (t) represents the brightness value of the i-th street light node at time t.

[0062] S12, collect traffic flow data in real time through traffic flow detectors arranged on the road i (t), where T i (t) represents the number of vehicles passing near the i-th street light node at time t.

[0063] S13, collect crowd density data P in real time through crowd flow sensors installed on street lights or in the surrounding environment i (t), where P i (t) represents the number of people flow near the i-th street light node at time t;

[0064] S14, collect the brightness data L i (t), traffic flow data T i (t) and crowd density data P i (t) is preprocessed to remove outliers and noise to form a structured environmental dataset D(t).

[0065]

[0066] Wherein, N represents the total number of street light nodes.

[0067] S20. Build an improved game theory model based on environmental data sets.

[0068] The present invention can transmit the collected environmental data set to a central control center for analysis and processing, and construct an improved game theory model in the central control center.

[0069] S30. Determine the optimal operation strategy for any street light node based on the improved game theory model.

[0070] The present invention can determine the best operation strategy of any street lamp node in the central control center based on the improved game theory model.

[0071] Most street light nodes in the existing system work independently and lack a coordination mechanism. Each street light node makes decisions based only on its own sensor data, and fails to fully utilize the information of adjacent nodes for collaborative optimization. This lack of collaborative working mode limits the performance of the overall system and makes it impossible to achieve global optimization.

[0072] In addition, the existing technology lacks effective anomaly detection and adaptive adjustment mechanisms. Due to the complex and changing environment, the street light system may encounter problems such as sensor failure or data anomalies in actual operation. Without effective anomaly detection and adaptive adjustment mechanisms, the system may continue to make decisions based on erroneous data in the event of a failure, resulting in system performance degradation and even safety hazards.

[0073] In summary, the existing technology still has the following defects: lack of coordination mechanism between nodes, lack of effective anomaly detection and adaptive adjustment mechanism.

[0074] As a possible implementation method, S30 determines the best operation strategy of any street light node based on an improved game theory model, including the following steps.

[0075] S31. Construct an improved game theory model and assign an initial strategy set S to each street light node i i , strategy set S i Including different lighting intensities L i (t) and switch state O i (t):

[0076] S i ={(L i1 ,O i1 ),(L i2 ,O i2 ),…,(L in ,O in )}

[0077] S32, dynamically adjust each strategy s according to the performance of each street light node i under different environmental data sets D(t) ij The weight w ij (t), where s ij ∈S i , weight w ij The adjustment of (t) is based on the feedback from the historical environment dataset and the changes in the real-time environment dataset.

[0078] S33. Based on the theory of dynamic alliance formation, street light nodes form temporary alliances according to real-time environmental data and their own status. Alliance members share information and resources and jointly formulate optimal strategies. Alliance C(t) is defined as the alliance set formed at time t. The cooperation benefit function U of each alliance member i and j is ij (t):

[0079] U ij (t) = β·E i (t)+γ·E j (t)-δ·C ij (t)

[0080] Among them, β and γ are the energy efficiency weights of node i and node j, E i (t) and E j (t) is the energy efficiency of node i and node j, C ij (t) is the collaboration cost between node i and node j, and δ is the collaboration cost weight.

[0081] S34. In high-traffic areas, street light nodes form an alliance C(t) to jointly adjust the lighting intensity to achieve the best lighting effect and energy-saving goals. The strategy optimization function π within each alliance c(t) :

[0082]

[0083] Among them, η C is the learning rate of the alliance, is the coalition policy gradient.

[0084] S35. Each street light node continuously optimizes its own strategy selection through adaptive learning. The learning process is based on the combination of reinforcement learning and deep learning technology. The node strategy optimization function π is defined. i (t).

[0085] S36. Embed an anomaly detection module in the construction environment dataset, monitor the status of each street light node in real time, define anomaly detection functions, and automatically adjust the strategy and send an alarm to the central control center when an anomaly is found.

[0086] Optionally, S32 dynamically adjusts each strategy s according to the performance of each street light node i under different environmental data sets D(t). ij The weight w ij (t), where s ij ∈S i , weight w ij The adjustment of (t) is based on the feedback of the historical environment dataset and the changes of the real-time environment dataset, and includes the following steps:

[0087] S321, initialize the strategy weight w of each street light node i ij (0), where s ij ∈S i Represents the jth strategy of the i-th street light node.

[0088] S322, at each time step t, according to the brightness data L in the environment data set D(t) i (t), traffic flow data T i (t) and crowd density data P i (t), calculate each strategy s ij The immediate return Rij (t):

[0089] R ij (t) = α L ·f(L i (t))+α T ·g(T i (t))+α P ·h(P i (t))

[0090] Among them, α L , α T , α P are the weight coefficients of brightness, traffic flow and crowd density, respectively, f(L i (t))、g(T i (t))、h(P i (t)) is the utility function of brightness, traffic flow and crowd density.

[0091] Brightness utility function f(L i (t)) Compare the current brightness value with the preset optimal brightness value:

[0092] f(L i (t))=-(L i (t)-L opt ) 2

[0093] Among them, L opt Indicates the optimal brightness value, L i (t) is the brightness value of the i-th street light node at time t. The brightness utility function represents the difference between the current brightness value and the optimal brightness value. The smaller the difference, the higher the utility.

[0094] Traffic flow utility function g(T i (t)) Compare the current traffic flow value with the preset optimal traffic flow value:

[0095]

[0096] Among them, T max represents the optimal traffic flow threshold, T i (t) is the traffic flow value near the i-th street light node at time t. The traffic flow utility function represents the ratio of the current traffic flow value to the optimal flow threshold. The closer the flow is to the optimal value, the higher the utility.

[0097] The utility function of crowd density h(P i (t)) Compare the current crowd density value with the preset optimal crowd density value:

[0098]

[0099] Among them, P max represents the optimal crowd density threshold, P i (t) is the crowd density value near the i-th street light node at time t. The crowd density utility function represents the ratio of the current crowd density value to the optimal density threshold. The closer the density is to the optimal value, the higher the utility.

[0100] S323, adjust each strategy s according to the feedback of the historical environment data set and the changes of the real-time environment data set ij The weight w ij (t), weight adjustment adopts the update rule in reinforcement learning:

[0101]

[0102] Among them, η1 is the learning rate, For strategies ij The expected return at time t.

[0103] S324, using the updated strategy weight w ij (t+1), calculate the optimal strategy s′ for each street light node according to the game theory model ij , the optimal strategy is chosen based on the strategy with the largest weight:

[0104]

[0105] in, is the optimal strategy for the i-th street light node at time t.

[0106] Optionally, each street light node in S35 continuously optimizes its own strategy selection through adaptive learning. The learning process is based on the combination of reinforcement learning and deep learning technology, and defines the node strategy optimization function π i (t), comprising the following sub-steps:

[0107] S351. Construct an adaptive learning model based on the combination of deep Q network and long short-term memory network.

[0108] S352, initialize the parameters of the deep Q network, define the state space S, action space A and reward function R, where the state space includes brightness data L i (t), traffic flow data T i (t) and crowd density data P i (t).

[0109] S353. Through the deep Q network, use the environmental data set D(t) for strategy learning and calculate the Q value Q(s,a) of each action a∈A.

[0110] S354. At each time step t, the environment data set D(t) is input into the long short-term memory network, the temporal features of the data are extracted, and the hidden state h(t) is output:

[0111] h(t)=LSTM(D(t))

[0112] S355. Combine the implicit state h(t) and the Q value Q(s,a) to calculate the comprehensive evaluation value V(a,t) of each action a:

[0113] V(a,t)=ω·Q(s,a)+λ·h(t)

[0114] Among them, ω and λ are weight coefficients.

[0115] S356. Select the optimal strategy a according to the comprehensive evaluation value V(a,t) * (t):

[0116] a * (t) = argmax a∈A V(a,t)

[0117] S357, each street light node according to the optimal strategy a * (t) Adjust its own strategy selection and continuously optimize the strategy. The learning process adopts experience replay and target network technology.

[0118] Optionally, S36 embeds an anomaly detection module in the construction environment data set, monitors the status of each street light node in real time, defines an anomaly detection function, and automatically adjusts the strategy and sends an alarm to the central control center when an anomaly is found, including the following sub-steps:

[0119] S361. Embed an anomaly detection module in the constructed environment data set D(t), and the anomaly detection module monitors the status of each street light node in real time.

[0120] S362. Define anomaly detection function A i (t), detect the abnormal state of each street light node, and the formula of the anomaly detection function is as follows:

[0121]

[0122] Where M is the number of monitoring parameters, is the current value of the kth parameter, is the historical mean of the kth parameter, and θ is the preset abnormal threshold.

[0123] S363. When an abnormal state is detected, the abnormal detection module automatically adjusts the strategy of the street light node, and the strategy adjustment is based on the result calculated by the improved game theory model.

[0124] S364. The anomaly detection module sends an alarm to the central control center. The alarm information includes the number of the abnormal street light node, the anomaly type and the current environmental data.

[0125] S40: Send the optimal operation strategy to the corresponding street light node, instructing the street light node to adjust the lighting intensity and / or switch state based on the optimal operation strategy.

[0126] The present invention can transmit the optimal operation strategy of each street light node calculated by the improved game theory model back to the respective street light nodes through the wireless communication network.

[0127] As a possible implementation, S40 instructs the street lamp node to adjust the lighting intensity and / or switch state based on the optimal operation strategy, including the following steps:

[0128] S41. Each street light node receives the optimal operation strategy transmitted by the central control center. Best operating strategy Including lighting intensity L i (t) and switch state O i (t):

[0129]

[0130] S42: According to the received operation strategy, the street light node adjusts the current lighting intensity to L i (t) and turn on or off the lighting device according to the specific situation. i (t):

[0131]

[0132] Among them, L i (t) represents the actual lighting intensity of the i-th street light node at time t, L max Indicates the maximum lighting intensity, setting the physical maximum lighting capacity of the street light node, α i Represents the adjustment coefficient, which is used to control the response degree of the current street light node to the optimal lighting intensity. represents the optimal lighting intensity calculated by the central control center for the i-th street light node at time t, β i represents the adjustment coefficient, which is used to control the current street light node’s dependence on the lighting intensity of adjacent nodes and reflects the degree of cooperation between nodes. N(i) represents the set of nodes adjacent to the i-th street light node, and w ij It represents the weight coefficient between node i and adjacent node j, which is used to measure the influence of adjacent node j on node i. The higher the weight, the greater the influence of the adjacent node on the current node. j(t-1) is the lighting intensity of the adjacent node j at time t-1, which represents the actual lighting intensity of the adjacent node at the previous moment and is used to calculate the lighting intensity adjustment at the current moment.

[0133] O i (t) represents the switch state of the i-th street light node at time t, 1 represents the on state, 0 represents the off state, γ i is the adjustment coefficient used to control the response degree of the current street light node to the optimal switch state. is the optimal switch state calculated by the central control center for the i-th street light node at time t, δ i is the adjustment coefficient used to control the degree of dependence of the current street light node on the switch state of the adjacent node, v ij is the switch state weight coefficient between node i and adjacent node j. The parameter is used to measure the influence of adjacent node j on the switch state of node i. j (t-1) is the switch state of the adjacent node j at time t-1, indicating the actual switch state of the adjacent node at the previous moment, which is used to calculate the switch state adjustment at the current moment. It is the threshold of the switch state, which is used to determine whether the switch state of the current node needs to be adjusted. When the comprehensive evaluation value exceeds the threshold, the node is turned on, otherwise it is turned off.

[0134] S43, street light nodes monitor their lighting status in real time and feedback the current lighting intensity L′ through built-in sensors i (t) and switch state O′ i (t):

[0135] L′ i (t) = f s (L i (t))

[0136] O′ i (t) = g s (O i (t))

[0137] Among them, f s and g s is the sensor feedback function, corresponding to the actual feedback values ​​of lighting intensity and switch status respectively.

[0138] S44, feedback data L' i (t) and O′ i (t) and expected value and Compare and if any difference is found, make adaptive adjustments. The adjustment process adopts closed-loop control strategy:

[0139]

[0140] Among them, ΔLi (t) and ΔO i (t) is the deviation value of the lighting intensity and switch state that needs to be adjusted.

[0141] S45, according to the deviation value ΔL i (t) and ΔO i (t), the street light node autonomously adjusts its lighting intensity and switch status to ensure that the actual lighting effect is consistent with the expected one:

[0142]

[0143] Among them, k L and k O are the adjustment coefficients of lighting intensity and switch state respectively, ξ and ζ are the influence coefficients of adjacent nodes, and Δt is the time step.

[0144] The present invention has the following beneficial effects.

[0145] (1) The present invention can collect and comprehensively analyze the environmental data of smart street lamp groups in real time by introducing Internet of Things technology and multiple sensors. In the central control center, an improved game theory model is applied to dynamically adjust the weight of each strategy according to the performance of each street lamp node under different environmental data sets, so that the system can make the optimal lighting adjustment according to the real-time environmental data, significantly improving the energy efficiency and responsiveness of the system.

[0146] (2) The present invention allocates an initial strategy set to each street light node, and through an improved game theory model, dynamically adjusts the weight of each strategy according to the changes in historical data and real-time data. By designing a collaborative competition mechanism, adjacent street light nodes cooperate in competition to ensure the overall lighting effect while maximizing the energy saving effect. Based on the theory of dynamic alliance formation, street light nodes can form temporary alliances according to real-time environmental data and their own status, share information and resources, and jointly formulate the best strategy. Through this collaborative competition mechanism, the system can achieve global optimization and improve the overall lighting effect and energy utilization efficiency.

[0147] (3) The present invention adopts an adaptive learning model that combines a deep Q network and a long short-term memory network, and combines reinforcement learning and deep learning techniques to continuously optimize the strategy selection of each street light node. By constructing an adaptive learning model, the system can continuously self-learn and adjust according to environmental changes to ensure that each street light node can make the best decision in different environments. The deep Q network is used to calculate the Q value of each action, and the long short-term memory network is used to extract the time series characteristics of the data. The two are combined to calculate the comprehensive evaluation value of each action, thereby selecting the optimal strategy. Through experience replay and target network technology, the learning stability and efficiency are improved.

[0148] The following is an explanation of an autonomous dispatching method for a street lamp group based on improved game theory provided by the present invention through an embodiment.

[0149] This example selects a main road in a large city, which is about 5 kilometers long and has 200 street light nodes along the way. The road has both busy traffic peaks and low traffic periods late at night. In order to verify the feasibility and effectiveness of the autonomous scheduling method of the smart street light group of the present invention, we conducted a one-month simulation experiment on the road and compared the energy consumption management and scheduling efficiency of the method of the present invention with the traditional street light control method.

[0150] In this embodiment, the traditional street lamp control method mainly relies on a fixed schedule for lighting scheduling, usually keeping all street lamps on from 7pm to 6am, without considering the changes in real-time traffic flow and pedestrian density. The method of the present invention, by introducing an improved game theory algorithm and an adaptive learning mechanism, collects environmental data in real time and performs intelligent scheduling based on these data.

[0151] First, light sensors, traffic flow detectors, and pedestrian flow sensors are installed on each street light node to collect brightness data, traffic flow data, and pedestrian density data in real time. These data are transmitted to the central control center through a wireless communication network to build an environmental data set.

[0152] In the central control center, an improved game theory model is applied to assign an initial strategy set to each street light node, including different lighting intensities and switch states. Through the feedback of historical data and the changes in real-time data, the weight of each strategy is dynamically adjusted, and a collaborative competition mechanism is designed to enable adjacent street light nodes to cooperate in competition, ensuring the overall lighting effect while maximizing energy saving.

[0153] An adaptive learning model based on a combination of a deep Q network and a long short-term memory network is used to continuously optimize the strategy selection of each street light node. Specifically, the deep Q network is used to calculate the Q value of each action, and the long short-term memory network is used to extract the temporal characteristics of the data. The two are combined to calculate the comprehensive evaluation value of each action, thereby selecting the optimal strategy.

[0154] An anomaly detection module is embedded in the environmental data set to monitor the status of each street light node in real time. When an anomaly is detected, the anomaly detection module automatically adjusts the strategy of the street light node and sends an alarm to the central control center. Through this mechanism, the system can maintain stable operation in a complex and changing environment.

[0155] To verify the feasibility of the method of the present invention, we conducted a one-month experiment on the road. The experimental data includes the brightness data of each street light node, traffic flow data and crowd density data.

[0156] During the evening rush hour on a certain day, the central control center received data from multiple street light nodes, showing a sudden increase in traffic flow. 45 "Data report, recorded brightness data L 45 (t) = 300 lux, traffic flow data T 45 (t) = 1200 vehicles / hour, passenger flow density data P 45 (t) = 300 people / hour. Based on these data, the system uses the game theory model to dynamically adjust the strategy and decides to “Node 45 The lighting intensity is increased to L 45 (t+1)=500lux, and the adjacent “Node 44 ” and “Node 46 ” Synchronously adjust to collaboratively increase brightness and ensure the safety of pedestrians and vehicles.

[0157] At 3 a.m., during the low period, the system detected that the traffic flow and pedestrian density at most street light nodes decreased significantly. 10 The data report shows the brightness data L 10 (t) = 200 lux, traffic flow data T 10 (t) = 50 vehicles / hour, passenger flow density data P 10 (t) = 20 people / hour. Based on this, the system decides to 10 The lighting intensity is reduced to L 10 (t+1)=100lux, turn off the adjacent "Node9" and "Node 11 "Some lighting equipment can be used to achieve energy saving.

[0158] On a weekend night, the system detected that the street light node "Node 87 The brightness data of " is abnormal. The specific data shows that L 87 (t) = 600 lux, which is much higher than the brightness level of other nodes, and the traffic flow of this node is T 87 (t) = 200 vehicles / hour and passenger density P 87 (t) = 50 people / hour does not support such a high brightness. The abnormal detection module triggers an alarm, records the abnormal data, and automatically sends the "Node 87 Adjust the brightness to a reasonable L 87 (t+1) = 300 lux. At the same time, the system sent an alarm to the central control center, reporting the abnormal node information and adjustment measures. The central control personnel then sent a maintenance team to check whether the sensor of the node was faulty and confirm the effect of the system's adaptive adjustment.

[0159] To demonstrate the effectiveness of the method of the present invention, we compared the energy consumption management and scheduling efficiency of the traditional method and the method of the present invention during the experiment. Table 1 is a comparison of experimental data between the traditional method and the method of the present invention.

[0160] Table 1 Comparison of experimental data between traditional method and the method of the present invention

[0161]

[0162] From the data in Table 1 above, it can be seen that within the same time period, energy consumption management: significantly reduces the energy consumption of the smart street light group, whether it is peak or trough period, it can effectively save energy, and the average energy consumption is saved by more than 30%. Fault response: greatly shortens the fault response time, improves the reliability and safety of the system, and increases the response speed by about 65%-68%. Dispatching efficiency: significantly improves the dispatching efficiency of the system, increases it by about 25%-28%, and optimizes the overall operation effect of the street light group. By improving the game theory algorithm, adaptive learning model and anomaly detection mechanism, the present invention not only reduces energy consumption, but also improves the response speed and dispatching efficiency of the system, providing solid technical support for the construction of smart cities.

[0163] Corresponding to the above-mentioned embodiment, the present invention further provides a street lamp group autonomous scheduling device based on improved game theory. Figure 3 A structural schematic diagram of a street light group autonomous scheduling device based on improved game theory is provided in an embodiment of the present invention. The device may include: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute a street light group autonomous scheduling method based on improved game theory as described in the above embodiment.

[0164] In a possible implementation of the present invention, the aforementioned at least one processor is capable of executing, through the Internet of Things technology, collecting brightness data, traffic flow data and pedestrian density data of any street light node in the smart street light group to construct an environmental data set; based on the environmental data set, constructing an improved game theory model; based on the improved game theory model, determining the optimal operating strategy for any street light node; sending the optimal operating strategy to the corresponding street light node, instructing the street light node to adjust the lighting intensity and / or switch status based on the optimal operating strategy.

[0165] In a specific implementation, the present invention can also be provided as a non-volatile computer storage medium, wherein the storage medium is a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium stores at least one program, each program includes instructions, and when the instructions are executed by a terminal, the terminal executes a street light group autonomous scheduling method based on improved game theory as described in the above embodiment.

[0166] In a possible implementation of the present invention, the aforementioned terminal executes, through the Internet of Things technology, to collect the brightness data, traffic flow data and pedestrian density data of any street light node in the smart street light group, and construct an environmental data set; based on the environmental data set, an improved game theory model is constructed; based on the improved game theory model, the optimal operation strategy of any street light node is determined; the optimal operation strategy is sent to the corresponding street light node, instructing the street light node to adjust the lighting intensity and / or switch status based on the optimal operation strategy.

[0167] The various embodiments of the present invention are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0168] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity 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, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0169] The above are only embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A street lamp group autonomous scheduling method based on improved game theory, characterized in that: The method comprises: Through the Internet of Things technology, the brightness data, traffic flow data and crowd density data of any street light node in the smart street light group are collected to build an environmental data set; Based on the environmental data set, construct an improved game theory model; Based on the improved game theory model, determining the optimal operation strategy of any street light node; The optimal operation strategy is sent to the corresponding street light node, instructing the street light node to adjust the lighting intensity and / or switch state based on the optimal operation strategy.

2. According to the improved game theory based autonomous dispatching method of street lamp groups according to claim 1, it is characterized in that: Based on the improved game theory model, determining the optimal operation strategy of any street light node includes: Based on the improved game theory model, an initial strategy set is allocated to any street lamp node, wherein any strategy in the initial strategy set includes different lighting intensities and switch states; Dynamically adjust the strategy weight of any strategy based on the performance of any street lamp node under different environmental data sets; Based on the strategy weight of any one of the strategies, the optimal operation strategy of any one of the street lamp nodes is determined.

3. The street lamp group autonomous scheduling method based on improved game theory according to claim 2 is characterized in that: Based on the performance of any street lamp node under different environmental data sets, dynamically adjusting the policy weight of any policy includes: At each time step t, calculating the instantaneous return of any of the strategies according to the brightness data, traffic flow data and crowd density data in the environmental data set; According to the formula Adjust the policy weight of any of the above policies, where w ij (t) is the strategy s ij The policy weight at time step t, R ij (t) is strategy s ij Immediate returns, For strategies ij The expected return at time t, η1 is the learning rate; Based on the strategy weight of any strategy, determining the best operation strategy of any street lamp node includes: making w ij The strategy with the largest (t+1) is determined as the optimal operation strategy for any street lamp node.

4. The street lamp group autonomous scheduling method based on improved game theory according to claim 2 is characterized in that: Based on the improved game theory model, determining the optimal operation strategy of any street lamp node also includes: Construct an adaptive learning model based on the combination of deep Q network and long short-term memory network; Initializing the parameters of the deep Q network, defining a state space, an action space, and a reward function, wherein the state space includes brightness data, traffic flow data, and crowd density data; Using the deep Q network, using the environment data set to perform strategy learning, and calculating the Q value of any action in the action space; At each time step t, the environment data set is input into the long short-term memory network, the temporal features of the environment data set are extracted, and the implicit state of the environment data set is output: Determining a comprehensive evaluation value of any one of the actions based on the Q value and the implicit state; Based on the comprehensive evaluation value, determining the optimal action in the action space; Instruct any street lamp node to adjust its own strategy selection according to the optimal action and continuously optimize the strategy. The learning process of the optimization strategy adopts experience playback and target network technology.

5. The street lamp group autonomous scheduling method based on improved game theory according to claim 2 is characterized in that: Based on the improved game theory model, determining the optimal operation strategy of any street lamp node also includes: Based on the dynamic alliance formation theory, the street light nodes in the smart street light group are instructed to form a temporary alliance according to the real-time environmental data and their own status, and the cooperation benefit function U of the two street light nodes in the temporary alliance is calculated. ij (t) = β·E i (t)+γ·E j (t)-δ·C ij (t), where β and γ are the energy efficiency weights of street light node i and street light node j, E i (t) and E j (t) is the energy efficiency of street light node i and street light node j, C ij (t) is the collaboration cost between streetlight node i and streetlight node j, δ is the collaboration cost weight; In high-traffic areas, the street light nodes in the smart street light group are instructed to form an alliance C(t) to jointly adjust the lighting intensity to achieve the best lighting effect and energy-saving goals. The strategy optimization function within each alliance is Among them, η C is the learning rate of the alliance C(t), is the coalition policy gradient.

6. The street lamp group autonomous scheduling method based on improved game theory according to claim 2 is characterized in that: Based on the improved game theory model, determining the optimal operation strategy of any street lamp node also includes: An anomaly detection module is embedded in the environmental data set, and the anomaly detection module is used to monitor the status of any street lamp node in real time; Calculate the abnormality detection function of any street light node, and when the abnormality detection function is greater than a preset abnormality threshold, determine that any street light node is in an abnormal state, and the abnormality detection function Where M is the number of monitoring parameters, is the current value of the kth parameter, is the historical mean of the kth parameter; When any of the street light nodes is detected to be in an abnormal state, the abnormality detection module is used to automatically adjust the strategy of the street light node; The anomaly detection module is also used to send alarm information to the central control center, and the alarm information includes the number of the street light node, the anomaly type and the current environment data set.

7. The street lamp group autonomous scheduling method based on improved game theory according to claim 1 is characterized in that: Instructing the street lamp node to adjust the lighting intensity and / or switch state based on the optimal operation strategy, including: Based on the formula Instruct any street lamp node to adjust the current lighting intensity to L based on the optimal operation strategy i (t), where L i (t) represents the actual lighting intensity of the i-th street light node at time t, L max Represents the maximum illumination intensity, α i Represents the adjustment coefficient used to control the response degree of the current street light node to the optimal lighting intensity. represents the optimal lighting intensity calculated by the i-th street light node at time t, β i represents the adjustment coefficient used to control the degree of dependence of the current street light node on the lighting intensity of adjacent nodes, N(i) represents the set of nodes adjacent to the i-th street light node, and w ij represents the weight coefficient between street light node i and its adjacent street light node j, L j (t-1) is the lighting intensity of the adjacent street lamp node j at time t-1; and / or, Based on the formula Instruct any street lamp node to adjust the current switch state to 0 based on the optimal operation strategy i (t), where O i (t) represents the switch state of the i-th street light node at time t, 1 represents the on state, 0 represents the off state, γ i is the adjustment coefficient used to control the response degree of the current street light node to the optimal switching state. is the optimal switch state calculated by the i-th street light node at time t, δ i is the adjustment coefficient used to control the degree of dependence of the current street light node on the switch state of the adjacent node, v ij is the switch state weight coefficient between street light node i and its adjacent street light node j, O i (t-1) is the switch state of adjacent node j at time t-1.

8. The street lamp group autonomous scheduling method based on improved game theory according to claim 1 is characterized in that: Instructing the street lamp node to adjust the lighting intensity and / or switch state based on the optimal operation strategy, further comprising: Based on the formula L′ i (t) = f s (L i (t)), obtain the feedback lighting intensity L′ of any street light node fed back by the lighting intensity sensor i (t), where f s is a sensor feedback function of the illumination intensity sensor; Calculate the feedback illumination intensity L′ i (t) and the expected value of the lighting intensity The lighting intensity deviation between i (t); Based on the formula L i (t+Δt)=L i (t)+k L ΔL i (t)+ξ·∑ j∈N(i) w ij ΔL j (t), instructing any street light node to adjust the lighting intensity, where k L is the adjustment coefficient of lighting intensity and switch state, ξ is the influence coefficient of adjacent nodes on lighting state adjustment; and / or, Based on the formula i (t) = g s (O i (t)), obtain the feedback switch state O′ of any street lamp node fed back by the switch state sensor i (t), where g s is a sensor feedback function of the switch state sensor; Calculate the feedback switch state O i (t) and expected value of switch state The switching state deviation between i (t); Based on the formula i (t+Δt)=O i (t)+k O ·ΔO i (t)+ζ·∑ j∈N(i) v ij ·ΔO j (t), instructing any street light node to adjust the switch state, where k O is the adjustment coefficient of the switch state, and ζ is the influence coefficient of the adjacent nodes on the switch state adjustment.

9. A street lamp group autonomous dispatching device based on improved game theory, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute the street lamp group autonomous scheduling method based on improved game theory according to any one of claims 1-8.

10. A non-volatile computer storage medium, characterized in that: The storage medium is a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium stores at least one program, each of which includes instructions, and when the instructions are executed by the terminal, the terminal executes a street light group autonomous scheduling method based on improved game theory according to any one of claims 1-8.

Images