Rail transit lighting energy-saving system based on wireless sensor network

Abstract

The application discloses a kind of track traffic lighting energy-saving systems based on wireless sensor network.The system includes network coordinator, at least two ultra-wideband anchor nodes, at least one vehicle-mounted label and at least one lighting terminal node.Network coordinator constructs and issues hybrid superframe structure, which is sequentially divided into positioning uplink time slot segment, control solving time slot segment and control downlink broadcast time slot segment in time domain.In the positioning uplink time slot segment, ultra-wideband anchor node performs ranging and reports original ranging data;In the control solving time slot segment, network coordinator calculates train position according to original ranging data and generates dimming instruction frame;In the control downlink broadcast time slot segment, network coordinator broadcasts dimming instruction frame.The application deeply integrates positioning service and control service by unified time slot scheduling, and realizes high-determinacy wireless control of track traffic platform lighting system.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication network technology and relates to an energy-saving system for rail transit lighting based on a wireless sensor network. Background Technology

[0002] The intelligent transformation of urban rail transit platform lighting systems requires the coordinated control of hundreds of lights over hundreds of meters of platform space, forming a moving light strip precisely synchronized with the train's position to achieve on-demand lighting and maximize energy savings. This necessitates a wireless sensor network to acquire the train's precise location in real time and deterministically distribute control commands to each light fixture.

[0003] Current solutions deploy train positioning and lighting control as separate subsystems. The positioning subnetwork typically employs wireless positioning methods based on received signal strength indication or independent ultra-wideband positioning protocols. The control subnetwork uses conventional wireless networks based on carrier sense multiple access mechanisms or independent industrial bus systems. Positioning information is relayed through a gateway or host computer before the control subnetwork sends dimming commands to the lighting fixtures.

[0004] The existing solution has two major drawbacks: First, the positioning and control subnetworks use different clock sources, communication protocols, and scheduling strategies, resulting in an unstable and uncontrollable system-level delay between sensing the train's position and executing lighting commands. This causes random phase lag between the lighting strip and the train's actual position, making it impossible to achieve precise light following the train's movement. Second, the medium access control layer of the control subnetwork is based on a random backoff mechanism and lacks the ability to synchronize the entire network's clock at the microsecond level. When controlling hundreds of nodes simultaneously, it cannot guarantee that all commands will be delivered and executed at the same fixed moment, leading to inconsistencies such as tearing and trailing of the lighting strip.

[0005] How to simultaneously carry high-reliability positioning and deterministic control services within the same wireless sensor network, and achieve strict time coupling between the two, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the problems existing in the background technology, this invention proposes an energy-saving lighting system for rail transit based on a wireless sensor network.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: a rail transit lighting energy-saving system based on a wireless sensor network, comprising a network coordinator, at least two ultra-wideband anchor nodes, at least one vehicle-mounted tag, and at least one lighting terminal node; The network coordinator constructs and publishes a hybrid superframe structure, which is sequentially divided in the time domain into a positioning uplink time slot segment, a control calculation time slot segment, and a control downlink broadcast time slot segment; During the positioning uplink time slot, the ultra-wideband anchor node performs ranging and reports the raw ranging data; During the control calculation time slot, the network coordinator acquires the raw ranging data, calculates the train position based on the raw ranging data, and generates a dimming command frame based on the train position. During the downlink broadcast time slot of the control, the network coordinator broadcasts the dimming command frame.

[0008] Specifically, ultra-wideband anchor nodes are pre-paired into anchor node pairs, each anchor node pair including a first anchor node and a second anchor node; Within one time slot of the positioning uplink time slot segment, the first anchor node sends a ranging initiation frame, the vehicle tag receives the ranging initiation frame, and sends a ranging response frame after a preset fixed response delay.

[0009] Specifically, the first anchor node receives the ranging response frame and calculates the first distance from the vehicle tag to the first anchor node based on the sending timestamp of the ranging initiation frame, the receiving timestamp and sending timestamp recorded by the vehicle tag, and the receiving timestamp of the ranging response frame. The second anchor node receives the ranging response frame and calculates the second distance from the vehicle tag to the second anchor node based on the preset fixed baseline distance with the first anchor node, the sending timestamp of the ranging initiation frame, and the receiving timestamp of the ranging response frame.

[0010] Specifically, the positioning uplink time slot segment includes multiple time slots, and multiple anchor node pairs are assigned to the multiple time slots; Among them, at least two anchor node pairs whose radio frequency coverage does not overlap are assigned to reuse the same time slot.

[0011] Specifically, the network coordinator includes a positioning engine that calculates the train's position based on the raw ranging data, specifically including: The positioning engine acquires the raw ranging values ​​and channel impulse response reports reported by each ultra-wideband anchor node; The positioning engine extracts the first-path power and total power from the channel impulse response report; The positioning engine calculates the ratio of initial path power to total power as the channel impulse response quality factor.

[0012] Specifically, the positioning engine performs the calculation of the train position based on the original ranging data, and further includes: the positioning engine inputs the original ranging value and the channel impulse response quality factor into an extended Kalman filter; In the update step of the extended Kalman filter, the positioning engine dynamically adjusts the observation noise covariance matrix according to the channel impulse response quality factor. The higher the channel impulse response quality factor, the smaller the corresponding observation noise variance setting value. The extended Kalman filter outputs the optimal state estimate of the train's position.

[0013] Specifically, the dimming instruction frame includes an absolute execution timestamp field and an illumination vector payload field; The absolute execution timestamp field stores a future specified time based on a unified clock across the entire network; The lighting vector load domain includes the mapping relationship between node identifiers and target brightness values; The lighting terminal node receives the dimming command frame, parses the absolute execution timestamp and its corresponding target brightness value, and performs a brightness refresh action when the local clock reaches the absolute execution timestamp.

[0014] Specifically, when generating the dimming instruction frame, the network coordinator reads the network time of the current superframe and sets the sum of the network time of the current superframe, one superframe period, and a preset protection margin as the absolute execution timestamp.

[0015] Specifically, the lighting terminal node includes a brightness register to be executed and an action trigger comparison register; The lighting terminal node loads the parsed target brightness value into the brightness register to be executed; The lighting terminal node loads the parsed absolute execution timestamp into the action trigger comparison register; The lighting terminal node continuously compares the count value of the local high-precision clock with the absolute execution timestamp in the action trigger comparison register. When the two match, a hardware interrupt is triggered to drive the output of the value of the brightness register to be executed to the lighting driver interface.

[0016] Specifically, the lighting drive interface is a Digital Addressable Lighting Interface (DALI-2) physical interface. When the hardware interrupt is triggered, the lighting terminal node converts the value of the brightness register to be executed into a PWM or analog drive signal and outputs it through the DALI-2 physical interface.

[0017] Compared with the prior art, the present invention has the following beneficial effects: by adopting a hybrid superframe structure that sequentially divides the time domain into positioning uplink time slots, control calculation time slots, and control downlink broadcast time slots, deterministic fusion scheduling of positioning services and control services within the same wireless sensor network is realized, eliminating the uncontrollable timing deviations caused by the separation of positioning and control networks in the existing scheme.

[0018] By implementing a time-division multiple access bidirectional double-hop ranging mechanism within the uplink time slot of the positioning system, and combining it with an extended Kalman filter that dynamically adjusts the observation noise covariance matrix based on the channel impulse response quality factor, the ranging cycle of the entire network with the same number of anchor nodes is shortened under dense anchor node deployment and strong multipath environment. It also effectively suppresses the contamination of positioning results by non-line-of-sight ranging outliers, and achieves high update rate and high accuracy train position estimation.

[0019] By introducing an absolute execution timestamp field into the dimming command frame, all lighting terminal nodes can synchronously execute brightness refresh actions at the same future moment defined by a unified clock across the entire network. This completely isolates the randomness of network transmission delay from the execution action, achieving sub-microsecond-level multi-node synchronous execution accuracy and ensuring tear-free and flicker-free movement of the light strip. Attached Figure Description

[0020] Figure 1 This is a block diagram of the overall system structure of the present invention; Figure 2 This is a schematic diagram of the overall system architecture and node deployment of the present invention; Figure 3 This is a timing diagram of the time division multiple access bidirectional double-hop ranging mechanism of the present invention; Figure 4 This is a schematic diagram of the dimming instruction frame structure with an absolute execution timestamp according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] like Figures 1-4 As shown, the technical solution adopted by this invention is as follows: A rail transit lighting energy-saving system based on a wireless sensor network includes a network coordinator, at least two ultra-wideband anchor nodes, at least one on-board tag, and at least one lighting terminal node. This system achieves highly deterministic closed-loop control from train position perception to synchronous execution of lighting commands by integrating the train positioning subnetwork and the lighting control subnetwork into a unified network architecture.

[0023] The network coordinator, the root node and sole clock source of the network, is deployed in the station equipment room. It builds and maintains the entire wireless sensor network, publishing a hybrid superframe structure across the network, divided in the time domain into uplink positioning time slots, control calculation time slots, and downlink control broadcast time slots. The network coordinator handles the computational tasks of train position calculation and control command generation, and broadcasts control commands containing synchronization time information to all lighting terminal nodes.

[0024] At least two ultra-wideband anchor nodes are fixedly deployed along the track to form a positioning sub-network. Each ultra-wideband anchor node is used to interact with the vehicle-mounted tag using ultra-wideband wireless signals, perform high-precision ranging, collect raw channel impulse response data reflecting the ranging quality, and report the raw ranging data and channel impulse response data to the network coordinator.

[0025] At least one vehicle-mounted tag is installed at the front or rear of the train as the target node to be located. The vehicle-mounted tag is used to receive ranging signals initiated by the ultra-wideband anchor node and reply with ranging response frames according to a preset fixed response delay, so as to cooperate with the ultra-wideband anchor node to complete the distance measurement of the vehicle-mounted tag.

[0026] At least one lighting terminal node is integrated into the LED luminaire driver power supply, forming a control sub-network. The lighting terminal node is used to receive dimming command frames broadcast by the network coordinator, parse the absolute execution timestamp and its corresponding target brightness value from the dimming command frame, and perform a brightness refresh action when the local clock reaches the absolute execution timestamp.

[0027] The coordinated work of the aforementioned nodes enables the train position information sensed by the positioning subnetwork to be processed by the network coordinator and then wirelessly and deterministically distributed to the control subnetwork with lighting control commands. This ensures that train position perception and lighting execution are strictly coupled in time, achieving precise and energy-saving lighting that follows the movement of the train.

[0028] In this technical solution, ultra-wideband anchor nodes are pre-paired into anchor node pairs, and each anchor node pair includes a first anchor node and a second anchor node.

[0029] The first anchor node is the anchor node that actively initiates ranging within the time slot allocated to the anchor node pair; the second anchor node is the anchor node in the same anchor node pair that only performs passive listening and completes ranging calculation. The physical conditions for pairing the first and second anchor nodes are that the vertical projection distance between the first and second anchor nodes in the track extension direction is less than 1 meter, and the horizontal distance between them is a known preset fixed baseline distance.

[0030] In one embodiment, the preset fixed baseline distance is set to 2 meters. The purpose of using anchor nodes in the deployment method is to support bidirectional double-hop ranging, so that the vehicle tag only needs to send one response frame, and both anchor nodes can obtain the distance to the vehicle tag.

[0031] At the start of an allocated time slot within a positioning uplink time slot segment of a hybrid superframe structure, the first anchor node transmits a ranging initiation frame within its radio frequency coverage area. This ranging initiation frame includes a frame type identifier, the node identifier of the first anchor node, and a transmission timestamp. The vehicle-mounted tag continuously listens within this time slot. Upon successfully receiving the ranging initiation frame, it records the reception timestamp. The vehicle-mounted tag does not immediately respond but transmits a ranging response frame after a preset fixed response delay. This fixed response delay is hardware-calibrated to ensure accuracy; in one embodiment, it is set to 128 microseconds.

[0032] The ranging response frame includes a frame type identifier, the node identifier of the vehicle tag, a sending timestamp of the ranging response frame, and a receiving timestamp of the ranging initiation frame recorded by the vehicle tag. The fixed response delay setting makes the entire ranging interaction time deterministic and known, eliminating ranging errors caused by the uncertainty of the vehicle tag's processing time.

[0033] The first anchor node receives the ranging response frame and records the receiving timestamp of the ranging response frame. Thus, the first anchor node obtains four timestamp data: the sending timestamp of the ranging initiation frame, the receiving timestamp of the ranging initiation frame recorded by the vehicle tag, the sending timestamp of the ranging response frame recorded by the vehicle tag, and the receiving timestamp of the ranging response frame.

[0034] The first anchor node calculates the distance from the vehicle-mounted tag to itself using the standard bilateral ranging method based on the four timestamps mentioned above, and defines this distance as the first distance. The calculation process involves: calculating the total round-trip time at the first anchor node, i.e., subtracting the sending timestamp of the ranging initiation frame from the receiving timestamp of the ranging response frame.

[0035] Calculate the internal processing delay of the vehicle-mounted tag by subtracting the receiving timestamp of the ranging initiation frame from the sending timestamp of the ranging response frame.

[0036] The total flight time of the signal in space is obtained by subtracting the internal processing delay of the vehicle-mounted tag from the total round-trip time.

[0037] Dividing the total round-trip flight time by two yields the one-way flight time of the signal between the first anchor node and the vehicle-mounted tag. Multiplying this one-way flight time by the speed of light gives the first distance. This calculation process achieves fully symmetrical bilateral bidirectional ranging, eliminating errors introduced by clock asynchrony between the transmitter and receiver.

[0038] The second anchor node in the same anchor node pair performs only passive listening within the same time slot and does not transmit any frames. The second anchor node receives the same ranging response frame emitted by the vehicle tag and records the reception timestamp of the ranging response frame. The second anchor node obtains the following parameters: the preset fixed baseline distance between itself and the first anchor node; the transmission timestamp of the ranging initiation frame obtained from the first anchor node through the wired bus; and the known fixed response delay experienced by the ranging response frame.

[0039] The second anchor node calculates the distance from the vehicle tag to itself based on the aforementioned parameters, defining it as the second distance. The calculation process is as follows: A time difference is calculated, equal to the timestamp of the receiving frame of the ranging response recorded by the second anchor node minus the timestamp of the sending frame of the ranging initiation frame shared by the first anchor node, and then minus the known fixed response delay. This time difference corresponds to the total propagation time of the signal from the first anchor node to the vehicle tag, and then from the vehicle tag to the second anchor node. Under the geometric constraint of a pre-set fixed baseline distance between the first and second anchor nodes, the second distance is calculated by solving the hyperbolic positioning equation.

[0040] In this mechanism, the vehicle tag responds only once. The second anchor node indirectly calculates its distance from the tag by using the known baseline length and precise time difference measurement, thus avoiding additional two-way interaction between the second anchor node and the tag.

[0041] Through the above mechanism, a single complete two-way frame interaction can obtain the first and second distances from the vehicle-mounted tag to the two anchor nodes. Compared with the traditional method where the tag must complete a complete two-way ranging with each anchor node one by one, this mechanism reduces the number of air frame interactions from four to two. When anchor nodes are densely deployed, it shortens the overall ranging cycle of the same number of anchor nodes by nearly 50%, effectively reducing air interface resource consumption and frame collision probability, and ensuring the positioning requirements for high update rates.

[0042] Furthermore, the network coordinator, as the root node and sole clock source of the network, is responsible for constructing and distributing the hybrid superframe structure across the entire network. This hybrid superframe structure is a periodic, repetitive time structure that deeply integrates location and control services in the time domain, and is the cornerstone for achieving deterministic communication in the network.

[0043] The hybrid superframe structure is sequentially divided into three functional segments in the time domain: the uplink positioning time slot segment, the control calculation time slot segment, and the downlink broadcast control time slot segment.

[0044] The positioning uplink time slot is a set of time slots in the hybrid superframe dedicated to ultra-wideband anchor nodes performing ranging and reporting raw ranging data to the network coordinator. Within this positioning uplink time slot, each ultra-wideband anchor node, according to the time slot allocation strategy, performs bidirectional, two-hop ranging interaction with the vehicle-mounted tag and reports the raw ranging values ​​and channel impulse response generated by the ranging.

[0045] The control calculation time slot immediately follows the positioning uplink time slot. During this control calculation time slot, the radio channel remains silent, and the positioning engine running in the network coordinator acquires the raw ranging data reported by all ultra-wideband anchor nodes. Based on this raw ranging data, it performs train position calculation and generates an optimal state estimate of the train's position. The control algorithm within the network coordinator then generates a dimming command frame containing the target brightness values ​​of each lighting terminal node based on this position estimate.

[0046] The control downlink broadcast time slot immediately follows the control solution time slot. During this control downlink broadcast time slot, the network coordinator transmits a wireless signal carrying a dimming command frame to all lighting terminal nodes via the control subnetwork broadcast channel.

[0047] In one embodiment, the superframe period of the hybrid superframe structure is set to 10 milliseconds, the total number of time slots is set to 10, and the basic length of each time slot is set to 1 millisecond. Specifically, the positioning uplink time slot occupies 8 time slots, the control calculation time slot occupies 1 time slot, and the control downlink broadcast time slot occupies 1 time slot. The 10-millisecond superframe period ensures that the lighting control command update rate is not less than 100 Hz, thereby meeting the smoothness requirements of the moving light strip.

[0048] Through the hybrid superframe structure that is sequentially divided in the time domain, the network coordinator completes the three stages of location data acquisition, control calculation, and command broadcasting in a conflict-free manner within a superframe period. This achieves deterministic fusion scheduling of location services and control services under the same network, eliminating uncontrollable timing deviations caused by the separation of the location network and the control network from the root.

[0049] Specifically, the positioning uplink time slot segment of the hybrid superframe structure comprises multiple time slots, the number of which is determined based on the number and spatial distribution of ultra-wideband anchor node pairs deployed in the network. In one embodiment, the positioning uplink time slot segment comprises eight time slots. All pre-paired anchor node pairs are assigned to the multiple time slots within this positioning uplink time slot segment, and each anchor node pair performs bidirectional, two-hop ranging interaction within its assigned specific time slot.

[0050] To achieve efficient utilization of spectrum resources, a spatial multiplexing time slot allocation strategy is adopted. When there are at least two anchor node pairs, and the radio frequency coverage areas of the at least two anchor node pairs do not overlap, the at least two anchor node pairs are allocated and multiplexed the same time slot.

[0051] The condition that the radio frequency coverage areas do not overlap is guaranteed by the spatial deployment spacing of the anchor node pairs: when the distance between physically non-adjacent anchor node pairs is greater than 50 meters, their radio frequency coverage areas do not overlap, and concurrent ranging operations in the same time slot will not cause mutual interference. Through this spatial multiplexing time slot allocation strategy, 14 anchor node pairs consisting of 28 ultra-wideband anchor nodes can be arranged in a time slot far less than the number of anchor node pairs to complete a complete ranging round, thereby ensuring the ranging update rate under high-density anchor node deployment within a limited superframe period.

[0052] Within the uplink time slot for positioning, each ultra-wideband anchor node in the allocated time slot performs ranging operations on the first and second anchor nodes it contains, according to the time-division multiple access bidirectional double-hop ranging procedure. The first and second anchor nodes each obtain the original ranging value with the vehicle tag.

[0053] Simultaneously, the ultra-wideband anchor node extracts the channel impulse response index from the physical layer pilot of the ranging response frame it receives. This channel impulse response index includes the first-path power and the total power of the received signal. Each ultra-wideband anchor node uses the raw ranging value and channel impulse response index generated by ranging as raw ranging data and reports it to the network coordinator via a wired bus (such as the CAN FD bus) for the positioning engine to perform train position calculation in subsequent control calculation time slots.

[0054] During the control calculation time slot, the network coordinator acquires the raw ranging data, calculates the train position based on the raw ranging data, and generates a dimming command frame based on the train position.

[0055] During the control and computation time slot of the hybrid superframe structure, the radio frequency channel remains silent, and the network coordinator obtains the raw ranging data reported by each ultra-wideband anchor node during the positioning uplink time slot through its internal wired data bus interface.

[0056] The raw ranging data includes: the raw ranging values ​​between each UWB anchor node and the vehicle tag obtained through the bidirectional double-hop ranging mechanism, and the channel impulse response report extracted by each UWB anchor node from the physical layer pilot of the received frame. The channel impulse response report includes the first path power and the total power of the received signal.

[0057] After the network coordinator acquires the raw ranging data, its internal positioning engine performs train position calculation based on this data. This calculation process includes: extracting the first-path power and total received signal power from the channel impulse response report; calculating the channel impulse response quality factor; inputting the raw ranging values ​​and the channel impulse response quality factor into an extended Kalman filter; dynamically adjusting the observation noise covariance matrix to perform weighted fusion of the ranging values; and outputting the optimal state estimate of the train position.

[0058] After obtaining the optimal state estimate of the train's position, the network coordinator's internal control algorithm calculates the target brightness value that each lighting terminal node should output in the next superframe cycle under the current train operating state, based on the train's current position and speed, combined with the physical layout information of the platform lighting fixtures. This target brightness value is encapsulated in the payload of the dimming command frame. The dimming command frame is a control command frame that will be broadcast to all lighting terminal nodes in the subsequent downlink broadcast time slot. Its structure includes an absolute execution timestamp field and a lighting vector payload field.

[0059] Through the acquisition, calculation, and generation pipeline operations implemented within a single node, the network coordinator tightly couples the location subnetwork and the control subnetwork in time and logic, eliminating uncontrollable timing deviations caused by the two networks belonging to different systems.

[0060] Specifically, the network coordinator includes a positioning engine. This positioning engine is a computational module running within the network coordinator, implementing its functionality as a software algorithm. Its function is to receive raw ranging data reported by each UWB anchor node and calculate the optimal state estimate of the train's position. Its input data includes: the raw ranging values ​​reported by each UWB anchor node, and the channel impulse response reports reported by each UWB anchor node.

[0061] The channel impulse response (COR) is a time-domain record of the wireless channel's response to an ultra-wideband (UWB) transmitted impulse signal. In the receiver of an UWB anchor node, the physical layer extracts the COR from the pilot symbols of the received ranging response frame. The COR report contains two key power parameters for evaluating ranging quality: first-path power, which is the energy value of the first path component of the received signal along the shortest line-of-sight path; and total received signal power, which is the total energy value of the received signal at the UWB anchor node, including all multipath reflection components.

[0062] After acquiring the raw ranging values ​​and channel impulse response reports reported by each ultra-wideband anchor node, the positioning engine extracts the first-path power and the total received signal power from each channel impulse response report. Then, the positioning engine performs a division operation, dividing the first-path power by the total received signal power, and the quotient is defined as the channel impulse response quality factor.

[0063] The physical meaning of the channel impulse response quality factor and its indicative role in ranging confidence are as follows: When the ratio of the first path power to the total power of the received signal approaches 1, it indicates that most of the energy of the received signal is concentrated on the first path component that arrives first along the shortest line-of-sight path, and the energy of the multipath reflection component is weak. The current measured original ranging value corresponds to the line-of-sight direct path with a high probability, and the ranging confidence is high. When the ratio of the first path power to the total power of the received signal is significantly less than 1, it indicates that the multipath reflection component dominates the energy of the received signal, the first path may be blocked or attenuated, and the current measured original ranging value is very likely to correspond to the non-line-of-sight reflection path, and the ranging confidence is low.

[0064] By constructing this channel impulse response quality factor, the positioning engine quantifies the channel quality information of the communication physical layer into a scalar index that can be directly used by the positioning algorithm, providing a basis for subsequent differentiated weighted fusion of different ranging values.

[0065] Specifically, the positioning engine calculates the train position based on the original ranging data, and further includes: the positioning engine inputting the original ranging value and the channel impulse response quality factor into an extended Kalman filter; in the update step of the extended Kalman filter, the positioning engine dynamically adjusts the observation noise covariance matrix according to the channel impulse response quality factor, the higher the channel impulse response quality factor, the smaller the corresponding observation noise variance setting value; the extended Kalman filter outputs the optimal state estimate of the train position.

[0066] The positioning engine internally operates an Extended Kalman Filter (EKF). The EKF is a recursive algorithm suitable for state estimation of nonlinear dynamic systems, its operation periodically alternating between two steps: a prediction step and an update step. In this system, the EKF's state vector consists of the train's position and velocity vectors, and the process model employs a uniformly accelerated motion model.

[0067] During state estimation, the positioning engine inputs the raw ranging values ​​reported by each ultra-wideband anchor node and the corresponding channel impulse response quality factor calculated by the positioning engine into the extended Kalman filter. The specific steps are as follows: First, the extended Kalman filter performs a prediction step, calculating the prior estimate of the train state for the current period based on the optimal state estimate and process model of the train position output from the previous period. Then, the extended Kalman filter enters the update step. In the update step, the filter needs to obtain the observation noise covariance matrix. The observation noise covariance matrix is ​​a diagonal matrix, where each element on the diagonal represents the noise variance of the corresponding ranging observation. The magnitude of this noise variance determines the weight of the ranging observation in updating the train state estimate: the smaller the noise variance, the higher the weight assigned to the observation; the larger the noise variance, the lower the weight assigned to the observation.

[0068] In this invention, the diagonal elements of the observation noise covariance matrix are not set as constants, but are dynamically adjusted based on the channel impulse response quality factor. The adjustment method is as follows: when the channel impulse response quality factor corresponding to a certain ultra-wideband anchor node is higher, it indicates a higher confidence level in the ranging value as a line-of-sight measurement. Therefore, the diagonal elements in the observation noise covariance matrix corresponding to that anchor node are set to smaller variance values, thus giving that ranging value a higher weight in the fusion process. Conversely, when the channel impulse response quality factor corresponding to a certain ultra-wideband anchor node is lower, it indicates a greater likelihood that the ranging value is a non-line-of-sight measurement. Therefore, the diagonal elements in the observation noise covariance matrix corresponding to that anchor node are set to larger variance values, thus automatically suppressing the contribution of that ranging value to state updates in the fusion process.

[0069] In one embodiment, the specific calculation method for the diagonal elements of the observation noise covariance matrix is ​​as follows: the i-th diagonal element in the observation noise covariance matrix, i.e., the observation noise variance corresponding to the i-th ultra-wideband anchor node, is set as the basic observation noise variance divided by the sum of the channel impulse response quality factor of that anchor node and a small positive value. This basic observation noise variance is a preset empirical constant value, and the small positive value is used to prevent division by zero errors when the channel impulse response quality factor is zero. The above calculation relationship can be accurately described by the following mathematical expression: .

[0070] in: The nth element in the observation noise covariance matrix represents the... The diagonal element, i.e., the th element The variance of observation noise corresponding to the ranging observations of each ultra-wideband anchor node; This represents the preset basic observation noise variance, which is an empirical constant. Indicates the first The channel impulse response quality factor corresponding to each ultra-wideband anchor node is calculated by dividing the first path power by the total power of the received signal. This represents a preset small positive constant used to prevent overflow when the denominator is zero.

[0071] After dynamically setting the observation noise covariance matrix, the extended Kalman filter calculates the Kalman gain and uses this gain to fuse all weighted ranging observations into the prior estimate of the train state, generating a posterior estimate of the train state. This posterior estimate is the optimal state estimate of the train position for the current cycle and is output by the extended Kalman filter for use by the control algorithm within the network coordinator to generate subsequent dimming command frames.

[0072] Through the above mechanism, the channel impulse response quality factor provided by the communication physical layer is seamlessly transformed into the observation confidence weight in the positioning algorithm, realizing cross-layer optimization design. This mechanism effectively suppresses the contamination of the positioning trajectory by outliers in non-line-of-sight ranging in the strong multipath environment of rail transit platforms, enabling the positioning engine to stably output high-precision, low-jitter optimal trajectory estimates of the train position when the train passes at high speed.

[0073] Within the control downlink broadcast time slot of the hybrid superframe structure, the network coordinator, as the only node in the network with broadcast privileges, wirelessly transmits the dimming command frame, which has been generated and encapsulated within the control calculation time slot, to all lighting terminal nodes via the broadcast channel of the control subnetwork. This control downlink broadcast time slot immediately follows the control calculation time slot, ensuring, in the time domain, that the pipelined operations from positioning data acquisition, position calculation, command generation to command issuance are completed sequentially and without conflict within one superframe cycle. During this time slot, all lighting terminal nodes are pre-scheduled to a wake-up receiving state, continuously listening for and receiving broadcast frames from the network coordinator.

[0074] The dimming command frame is generated by the network coordinator within the control calculation time slot. In addition to the regular frame header and frame check sequence, its frame structure contains a payload consisting of two core fields: the absolute execution timestamp field and the illumination vector payload field.

[0075] The absolute execution timestamp field stores a specified future time based on a unified network clock. This unified network clock is established by the network coordinator as the root clock source and continuously distributed for calibration to all network nodes. All lighting terminal nodes maintain microsecond-level synchronization accuracy with this unified network clock. The specified future time stored in the absolute execution timestamp field is an absolute time value, meaning that all lighting terminal nodes should and should only execute the brightness refresh action specified in this frame at the instant their local clock count precisely matches this time, regardless of any deviation in the actual time each node receives the frame.

[0076] The lighting vector payload domain stores a mapping table between node identifiers and target brightness values. This mapping table consists of a set of entries, each containing the node identifier of a lighting terminal node and the target brightness value that node should output within the current superframe period. In one embodiment, the mapping table contains 200 entries, covering all lighting terminal nodes on the platform, with the target brightness value resolution accurate to 0.1% dimming precision, following the logarithmic dimming curve of the digitally addressable lighting interface.

[0077] The lighting terminal node wakes up within its assigned broadcast receive time slot and receives a dimming command frame broadcast by the network coordinator. The lighting terminal node parses the dimming command frame, extracting two key pieces of information: first, the absolute execution timestamp parsed from the absolute execution timestamp field; and second, the target brightness value corresponding to itself, found in the mapping table of the lighting vector payload field based on its own node identifier. After parsing this information, the lighting terminal node does not immediately perform brightness adjustment. Instead, it loads the target brightness value into its local pending brightness register and the absolute execution timestamp into its local action trigger comparison register. The lighting terminal node's internal high-precision clock runs continuously, its count value constantly compared with the absolute execution timestamp stored in the action trigger comparison register. The instant the local clock count value precisely matches the absolute execution timestamp, a hardware interrupt is triggered, driving the lighting terminal node to perform a brightness refresh action and output the target brightness value stored in the pending brightness register to the lighting driver interface.

[0078] This mechanism completely isolates the randomness of network transmission delay from the execution action. The dimming action of all lighting terminal nodes is locked at the same absolute time point defined by the unified clock of the whole network, realizing the multi-node synchronous execution accuracy at the sub-microsecond level. It is the fundamental technical means to ensure that the mobile light strip is tear-free and flicker-free.

[0079] When the network coordinator generates a dimming command frame within the control calculation time slot, it performs the following steps to determine the value to be filled in the absolute execution timestamp field. The network coordinator reads the network time of the current superframe maintained by its local time synchronization channel frequency hopping protocol stack. This current superframe network time is the absolute time count value at the start time of the current hybrid superframe. Subsequently, the network coordinator performs an addition operation, adding the read current superframe network time, the duration of one superframe period, and a preset protection margin. The resulting sum is set as the absolute execution timestamp and written into the absolute execution timestamp field of the dimming command frame.

[0080] The preset protection margin is a positive time constant. Its function is to compensate for the control calculation time and frame transmission delay in the worst case, ensuring that all lighting terminal nodes have successfully received and parsed the dimming instruction frame and completed the loading operation of the brightness register to be executed and the action trigger comparison register before the time specified by the absolute execution timestamp arrives.

[0081] In one embodiment, the superframe period is set to 10 milliseconds, and the preset protection margin is set to 500 microseconds. Under this configuration, if the network time of the current superframe is a certain value, the absolute execution timestamp is set to the network time of the current superframe plus 10.5 milliseconds. This setting ensures that the future time specified by the absolute execution timestamp is approximately 500 microseconds after the start of the next superframe period, at which point all lighting terminal nodes will synchronously perform a brightness refresh operation.

[0082] Specifically, the lighting terminal node includes two dedicated registers at the hardware level: a brightness register to be executed and an action trigger comparison register.

[0083] The pending brightness register is used to temporarily store the target brightness value corresponding to this lighting terminal node, which is parsed from the received dimming command frame. This target brightness value is stored in digital form and represents the target brightness level that this node should output in the current dimming cycle.

[0084] The action trigger compare register stores the absolute execution timestamp parsed from the absolute execution timestamp field of the dimming instruction frame. This absolute execution timestamp is a count value based on a future specified time using a unified network clock.

[0085] After successfully receiving the dimming command frame broadcast by the network coordinator during the downlink broadcast time slot, the lighting terminal node executes the following hardware operation procedure: The first step is for the lighting terminal node to find its own target brightness value from the lighting vector payload field of the dimming command frame based on its own node identifier, and then load the digital value of the target brightness value into the brightness register to be executed. The second step is for the lighting terminal node to parse the absolute execution timestamp from the absolute execution timestamp field of the dimming instruction frame and load the count value of the absolute execution timestamp into the action trigger comparison register.

[0086] After the loading operation is completed, the local high-precision clock inside the lighting terminal node continues to run freely. This local high-precision clock is continuously calibrated by the network coordinator through the time synchronization channel frequency hopping protocol stack, maintaining microsecond-level synchronization accuracy with the unified clock of the entire network. In one embodiment, this local high-precision clock is driven by a 32.768 kHz temperature-compensated crystal oscillator, with a resolution of approximately 30.5 microseconds.

[0087] The hardware comparison logic of the lighting terminal node continuously compares the current count value of its local high-precision clock with the absolute execution timestamp stored in the action trigger comparison register cycle by cycle. The moment the local high-precision clock count value precisely matches the absolute execution timestamp in the action trigger comparison register, a hardware interrupt is triggered. This hardware interrupt drives the lighting terminal node to perform a brightness refresh action, outputting the value stored in the brightness register to the lighting driver interface.

[0088] Through the aforementioned hardware registers and interrupt mechanisms, when all lighting terminal nodes perform brightness refresh actions, the timing of their actions is strictly triggered by the local hardware comparator based on the absolute execution timestamp. The randomness of network transmission delay is completely isolated from the execution actions, and the synchronization execution accuracy between nodes reaches the sub-microsecond level.

[0089] Specifically, the lighting driver interface is the physical connection interface between the lighting terminal node and the LED luminaire driver power supply. This lighting driver interface specifically adopts the Digital Addressable Lighting Interface (DALI-2) physical interface. DALI-2 is a digital lighting control protocol defined by the International Electrotechnical Commission (IEC) standard 62386 series. Its physical interface specifies electrical characteristics, signal levels, and communication frame formats, supporting independent addressing and precise dimming control for each luminaire. Using this standard interface allows the lighting terminal node of this system to be directly compatible with LED driver power supplies that conform to the DALI-2 standard.

[0090] When a hardware interrupt is triggered, the lighting terminal node performs signal conversion and output operations. The lighting terminal node reads the digital target brightness value stored in the brightness register to be executed and converts this digital target brightness value into a pulse width modulation (PWM) signal or an analog drive signal. This conversion process is completed by the digital-to-analog converter circuit or PWM generation module inside the lighting terminal node. The converted PWM signal or analog drive signal is output to the dimming input of the LED luminaire driver power supply through the electrical connection line of the DALI-2 digital addressable lighting interface, thereby controlling the LED luminaire to refresh its brightness according to the target brightness value.

[0091] In one embodiment, the resolution of the target brightness value in the dimming command frame is accurate to 0.1 percent, following the logarithmic dimming curve defined by the Digital Addressable Lighting Interface (DALI-2) standard, in order to adapt to the non-linear perception characteristics of the human eye to brightness changes and achieve a visually smooth and flicker-free moving light strip effect.

[0092] In one specific embodiment, a 200-meter-long island platform is used as the application scenario. 200 LED lights, numbered L1 to L200, are installed at 1-meter intervals along the platform edge. Each LED light's driver integrates a Time Synchronization Channel Frequency Hopping (TSCH) wireless communication module conforming to the IEEE 802.15.4e standard, making it a lighting terminal node in the network. Along the railings on both sides of the platform tracks, 28 ultra-wideband anchor nodes, numbered A1 to A28, are installed at 15-meter intervals. An onboard tag is installed at the bottom of the front of each operating train. The network coordinator is deployed in the platform equipment room.

[0093] After the system powers on, the network coordinator, acting as the root node and the sole clock source, begins broadcasting enhanced beacon frames. All UWB anchor nodes and illumination terminal nodes listen for the beacons on their common channel, completing coarse clock synchronization with the network coordinator and the TSCH network entry procedure. Upon successful network entry, the network coordinator assigns a unique 16-bit short address to each node and publishes the hybrid superframe structure configuration to the entire network. The superframe cycle of the hybrid superframe structure... Set to 10 milliseconds, total number of time slots Set to 10, with a base length for each time slot. Set to 1 millisecond. The hybrid superframe structure is sequentially divided into three functional segments in the time domain: the uplink positioning time slot segment occupies 8 time slots from Slot 0 to Slot 7, the control and calculation time slot segment occupies 1 time slot (Slot 8), and the downlink broadcast control time slot segment occupies 1 time slot (Slot 9). Superframe period Setting it to 10 milliseconds ensures that the update rate of lighting control commands is no less than 100 Hz, meeting the smoothness requirements of the moving light strip. This parameter is chosen based on the fact that the human eye's persistence of vision for dynamic light strip changes requires a refresh rate of no less than 100 Hz; this value is a typical engineering experimental value in the display and lighting control field.

[0094] During the superframe cycle For 10 milliseconds, the total number of time slots In a hybrid superframe structure with 10 nodes, network time is divided into continuous, periodically repeating time units, and the transceiver state machines of all nodes are strictly controlled by their local time slot scheduling tables. Within one superframe cycle, the system sequentially executes three stages of operations: uplink positioning, control calculation, and downlink control broadcasting.

[0095] During the positioning uplink phase, i.e., during the execution of Slots 0 to 7, the 28 ultra-wideband anchor nodes were pre-paired into 14 anchor node pairs. Each anchor node pair consists of a first anchor node and a second anchor node. The pairing deployment conditions for the first and second anchor nodes are: the vertical projection distance between them in the track extension direction is less than 1 meter, and the horizontal distance between them is the known preset fixed baseline distance. In this embodiment The value is set to 2 meters. The preset fixed baseline distance is... Setting the distance to 2 meters is a typical engineering experimental value for the installation of trackside equipment on rail transit platforms. This spacing ensures that the two anchor nodes have sufficient spatial diversity gain to distinguish multipaths, while also preventing the distance from exceeding the reliable data sharing distance of the wired bus due to excessive spacing.

[0096] Fourteen anchor node pairs are allocated to eight time slots in the positioning uplink time slot segment. When the distance between physically non-adjacent anchor node pairs is greater than 50 meters, their radio frequency coverage areas do not overlap. In this embodiment, anchor node pairs located at both ends of the platform (such as A1 / A2 and A15 / A16 pairs) are allocated and multiplexed to the same time slot, performing ranging concurrently within Slot 0 without interference. Through this spatial multiplexing time slot allocation strategy, 28 ultra-wideband anchor nodes only require eight time slots to complete a full ranging cycle, ensuring the 10-millisecond superframe period is achieved.

[0097] Taking anchor node pair A1 / A2 located at the beginning of the platform as an example, this illustrates the time-division multiple access bidirectional double-hop ranging procedure executed within one time slot of the positioning uplink time slot segment. At the start of Slot 0, the first anchor node A1 transmits a ranging initiation frame within its radio frequency coverage area. The ranging initiation frame includes a frame type identifier, the node identifier of A1, and a transmission timestamp of the ranging initiation frame. The vehicle-mounted tag continuously listens within Slot 0. After successfully receiving a ranging initiation frame, it records the timestamp of the received ranging initiation frame. The vehicle-mounted tag experiences a preset fixed response delay. Then, a ranging response frame is sent. In this embodiment, a fixed response delay is used. The value is 128 microseconds. This delay is ensured for accuracy through hardware calibration. 128 microseconds is a typical engineering experimental value for ultra-wideband transponder hardware processing, sufficient to accommodate the time required for physical layer preamble detection, frame synchronization, and data demodulation. The ranging response frame includes a frame type identifier, the node identifier of the vehicle tag, and a timestamp of the ranging response frame's transmission. And the timestamp of the ranging initiation frame recorded by the vehicle tag. .

[0098] The first anchor node A1 receives the ranging response frame and records the receiving timestamp of the ranging response frame. The first anchor node A1 obtains four timestamp data: the sending timestamp of the ranging initiation frame. The timestamp of the ranging initiation frame recorded by the vehicle tag. The transmission timestamp of the ranging response frame recorded by the vehicle tag. and the received timestamp of the ranging response frame Based on the four timestamps mentioned above, the first anchor node A1 calculates the first distance from the vehicle tag to the first anchor node A1 using a bilateral ranging method. The calculation process is as follows: calculate the total round-trip time at the first anchor node, which is the timestamp of the ranging response frame. Subtract the transmission timestamp of the ranging initiation frame ; Calculate the internal processing latency of the vehicle-mounted tag, i.e., the timestamp of the ranging response frame transmission. Subtract the received timestamp of the ranging initiation frame Subtract the internal processing delay of the vehicle-mounted tag from the total round-trip time to obtain the total flight time of the signal in space; divide the total round-trip flight time by two to obtain the one-way flight time of the signal between the first anchor node A1 and the vehicle-mounted tag; multiply the one-way flight time by the speed of light. The resulting product is the first distance. .

[0099] In the same anchor node pair, the second anchor node A2 only performs passive listening within the same Slot0 and does not send any frames. The second anchor node A2 receives the same ranging response frame from the vehicle-mounted tag and records the receiving timestamp of the ranging response frame. The second anchor node A2 acquires the following parameters: the preset fixed baseline distance between it and the first anchor node A1. (Value is 2 meters); the transmission timestamp of the ranging initiation frame obtained from the first anchor node A1 via the CAN FD wired bus. ; and known fixed response delay (Value is 128 microseconds).

[0100] The second anchor node A2 calculates the second distance from the vehicle tag to the second anchor node A2 based on the above parameters. The calculation process is as follows: Calculate a time difference, which is equal to the received timestamp of the ranging response frame. Subtract the transmission timestamp of the ranging initiation frame Subtract the known fixed response delay. This time difference corresponds to the total propagation time of the signal from the first anchor node A1 to the vehicle tag, and then from the vehicle tag to the second anchor node A2; given a known preset fixed baseline distance. Under the geometric constraints, the second distance is calculated by solving the hyperbola positioning equation. .

[0101] After Slot 0 is completed, the remaining anchor nodes perform ranging sequentially or concurrently within their respective allocated time slots in Slots 1 through 7, in the same manner. After completing the ranging, each ultra-wideband anchor node packages and reports the original ranging value, along with the channel impulse response index extracted from the physical layer pilot of the ranging response frame, to the network coordinator via the CAN FD bus. The channel impulse response index includes the first-path power. and total power of received signal .

[0102] During the control calculation phase, specifically during Slot 8 execution, the radio channel remains silent. The network coordinator acquires raw ranging data reported by all UWB anchor nodes via the CAN FD bus. The positioning engine running internally within the network coordinator performs train position calculation. The positioning engine is a software-based computational module that takes as input the raw ranging values ​​and channel impulse response reports reported by each UWB anchor node, and outputs an optimal state estimate of the train's position.

[0103] The positioning engine extracts the first-path power from each channel impulse response report. and total power of received signal Calculate the power of the first diameter Total power of received signal The ratio of the two values ​​is used to obtain the channel impulse response quality factor corresponding to the ultra-wideband anchor node. When the channel impulse response quality factor When the value approaches 1, it indicates that most of the received signal energy is concentrated in the first path component of the shortest line-of-sight path, and the confidence level of this original ranging value is high; when the channel impulse response quality factor... When the value is significantly less than 1, it indicates that the multipath reflection component energy is dominant, and the confidence level of the original ranging value is low.

[0104] The positioning engine internally operates an extended Kalman filter. The state vector of the extended Kalman filter consists of the train's position and velocity vectors, and the process model adopts a uniformly accelerated motion model. The positioning engine uses the raw ranging values ​​from 28 ultra-wideband anchor nodes and their corresponding channel impulse response quality factors. Input extended Kalman filter.

[0105] The Extended Kalman Filter (EKF) first performs a prediction step, calculating a priori estimates of the train's state for the current cycle based on the optimal state estimate and process model of the train's position output from the previous cycle. Then, it proceeds to the update step, where the EKF needs to construct the observation noise covariance matrix. Observation noise covariance matrix For one A diagonal square matrix, the first element on its diagonal... element That is, the first The observation noise variance corresponding to each ultra-wideband anchor node is set as the basic observation noise variance. Divide by the channel impulse response quality factor of the anchor node With a small positive constant The sum. In this embodiment, the variance of the basic observation noise. The preset value is 0.01 square meters, a small positive constant. The value is 0.001, and these two values ​​are typical engineering experimental values ​​for the Kalman filter tuning parameters. The observation noise covariance matrix is ​​constructed in this way. In the middle, the channel impulse response quality factor The higher the anchor node, the higher its corresponding diagonal element. The smaller the value, the higher the weight of the original ranging value of the anchor node in the subsequent Kalman gain calculation; Channel impulse response quality factor The lower the anchor node, the more diagonal elements it corresponds to. The larger the value, the more the contribution of the original ranging value of the anchor node to the state update is automatically suppressed. The extended Kalman filter completes the observation noise covariance matrix... After dynamic setting, the Kalman gain is calculated, and the Kalman gain is used to fuse all weighted observations into the prior estimate of the train state to generate the posterior estimate of the train state, that is, the optimal state estimate of the train position in this cycle.

[0106] After obtaining the optimal state estimate of the train's position, the network coordinator's internal control algorithm calculates the target brightness value that each lighting terminal node should output in the next superframe cycle based on the train's current position and speed, combined with the physical layout information of the 200 LED lights on the platform. The target brightness value is encapsulated to generate a dimming command frame. In the frame structure of the dimming command frame, besides the frame header and frame check sequence, the payload consists of an absolute execution timestamp field and a lighting vector payload field. The network coordinator reads the network time of the current superframe from the current TSCH protocol stack. The absolute execution timestamp Set as the network time of the current superframe Add superframe cycle (10 milliseconds) plus preset protection margin (500 microseconds). Preset protection margin. The value of 500 microseconds is an engineering experimental value representing the sum of the maximum air transmission delay and frame processing time typical in ultra-wideband networks. The lighting vector load domain contains 200 entries, each an array of node identifiers mapped to target brightness values. The target brightness values ​​have a dimming accuracy of 0.1 percent, conforming to the logarithmic dimming curves of digitally addressable lighting interfaces defined by the International Electrotechnical Commission (IEC) standard 62386 series.

[0107] During the downlink broadcast phase of control, i.e., during the execution of Slot 9, the network coordinator sends dimming command frames through the control subnetwork broadcast channel. All lighting terminal nodes L1 to L200 are awakened and receive the frames within Slot 9. Taking lighting terminal node L1 as an example, after receiving the dimming command frame, lighting terminal node L1 looks up the corresponding target brightness value from the lighting vector payload field based on its own node identifier, loads the target brightness value into the brightness register to be executed, and parses the absolute execution timestamp from the absolute execution timestamp field. The absolute execution timestamp Load the action trigger compare register. The local high-precision clock inside the lighting terminal node L1 is driven by a 32.768 kHz temperature-compensated crystal oscillator with a resolution of approximately 30.5 microseconds. This clock is continuously calibrated by the network coordinator via the TSCH protocol stack. The hardware compare logic of the lighting terminal node L1 continuously compares the current count value of the local high-precision clock with the absolute execution timestamp in the action trigger compare register. Compare them.

[0108] When the local high-precision clock count value is compared with the absolute execution timestamp At the precise moment of matching, a hardware interrupt is triggered. The hardware interrupt drives lighting terminal node L1 to convert the digital target brightness value in the brightness register to be executed into a pulse width modulation signal, which is then output to the dimming input of the LED luminaire driver power supply via the Digital Addressable Lighting Interface (DALI-2) physical interface. Lighting terminal nodes L1 to L200 execute at the same absolute timestamp. The brightness refresh action is executed simultaneously at the same time.

[0109] Through the above superframe period With a continuous cycle of 10 milliseconds, the system stably completes the closed loop of perception, decision-making, and execution at an update rate of 100 Hz. In the environment of high-speed trains passing through strong multipath platforms, the optimal state estimation accuracy of the train position output by the positioning engine reaches ±10 cm, the synchronous dimming execution accuracy of 200 LED lights reaches the sub-microsecond level, and the energy consumption of ineffective lighting is reduced by more than 85% when a single train enters the station.

[0110] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A rail transit lighting energy-saving system based on a wireless sensor network, characterized in that, It includes a network coordinator, at least two ultra-wideband anchor nodes, at least one vehicle-mounted tag, and at least one lighting terminal node; The network coordinator constructs and publishes a hybrid superframe structure, which is sequentially divided in the time domain into a positioning uplink time slot segment, a control calculation time slot segment, and a control downlink broadcast time slot segment; During the positioning uplink time slot, the ultra-wideband anchor node performs ranging and reports the raw ranging data; During the control calculation time slot, the network coordinator acquires the raw ranging data, calculates the train position based on the raw ranging data, and generates a dimming command frame based on the train position. During the downlink broadcast time slot of the control, the network coordinator broadcasts the dimming command frame.

2. The energy-saving rail transit lighting system based on a wireless sensor network according to claim 1, characterized in that, Ultra-wideband anchor nodes are pre-paired into anchor node pairs, each anchor node pair including a first anchor node and a second anchor node; Within one time slot of the positioning uplink time slot segment, the first anchor node sends a ranging initiation frame, the vehicle tag receives the ranging initiation frame, and sends a ranging response frame after a preset fixed response delay.

3. The energy-saving rail transit lighting system based on a wireless sensor network according to claim 2, characterized in that, The first anchor node receives the ranging response frame and calculates the first distance from the vehicle tag to the first anchor node based on the sending timestamp of the ranging initiation frame, the receiving timestamp and sending timestamp recorded by the vehicle tag, and the receiving timestamp of the ranging response frame. The second anchor node receives the ranging response frame and calculates the second distance from the vehicle tag to the second anchor node based on the preset fixed baseline distance with the first anchor node, the sending timestamp of the ranging initiation frame, and the receiving timestamp of the ranging response frame.

4. The energy-saving rail transit lighting system based on a wireless sensor network according to claim 1, characterized in that, The positioning uplink time slot segment includes multiple time slots, and multiple anchor node pairs are assigned to the multiple time slots; Among them, at least two anchor node pairs whose radio frequency coverage does not overlap are assigned to reuse the same time slot.

5. A rail transit lighting energy-saving system based on a wireless sensor network according to claim 1, characterized in that, The network coordinator includes a positioning engine, which performs the calculation of the train's position based on the raw ranging data, specifically including: The positioning engine acquires the raw ranging values ​​and channel impulse response reports reported by each ultra-wideband anchor node; The positioning engine extracts the first-path power and total power from the channel impulse response report; The positioning engine calculates the ratio of initial path power to total power as the channel impulse response quality factor.

6. A rail transit lighting energy-saving system based on a wireless sensor network according to claim 5, characterized in that, The positioning engine performs the calculation of the train position based on the original ranging data, and further includes: the positioning engine inputs the original ranging value and the channel impulse response quality factor into an extended Kalman filter; In the update step of the extended Kalman filter, the positioning engine dynamically adjusts the observation noise covariance matrix according to the channel impulse response quality factor. The higher the channel impulse response quality factor, the smaller the corresponding observation noise variance setting value. The extended Kalman filter outputs the optimal state estimate of the train's position.

7. A rail transit lighting energy-saving system based on a wireless sensor network according to claim 1, characterized in that, The dimming instruction frame includes an absolute execution timestamp field and an illumination vector payload field; The absolute execution timestamp field stores a future specified time based on a unified clock across the entire network; The lighting vector load domain includes the mapping relationship between node identifiers and target brightness values; The lighting terminal node receives the dimming command frame, parses the absolute execution timestamp and its corresponding target brightness value, and performs a brightness refresh action when the local clock reaches the absolute execution timestamp.

8. A rail transit lighting energy-saving system based on a wireless sensor network according to claim 7, characterized in that, When generating the dimming instruction frame, the network coordinator reads the network time of the current superframe and sets the sum of the network time of the current superframe, one superframe period, and a preset protection margin as the absolute execution timestamp.

9. A rail transit lighting energy-saving system based on a wireless sensor network according to claim 7, characterized in that, The lighting terminal node includes a brightness register to be executed and an action trigger comparison register; The lighting terminal node loads the parsed target brightness value into the brightness register to be executed; The lighting terminal node loads the parsed absolute execution timestamp into the action trigger comparison register; The lighting terminal node continuously compares the count value of the local high-precision clock with the absolute execution timestamp in the action trigger comparison register. When the two match, a hardware interrupt is triggered to drive the output of the value of the brightness register to be executed to the lighting driver interface.

10. A rail transit lighting energy-saving system based on a wireless sensor network according to claim 9, characterized in that, The lighting drive interface is a Digital Addressable Lighting Interface (DALI-2) physical interface. When the hardware interrupt is triggered, the lighting terminal node converts the value of the brightness register to be executed into a PWM or analog drive signal and outputs it through the DALI-2 physical interface.

Images