During high-speed train operation, the temperature of the bearings in the running gear continuously rises. Excessive axle temperature can cause problems such as overheating and axle shedding, seriously affecting railway transportation safety. Currently, two main types of axle temperature detection devices are used on Chinese railway trains: a direct-contact onboard axle temperature detection system and a non-contact infrared axle temperature detection system. The former uses temperature sensors installed in the axle box to collect temperature data, which is then transmitted via cable to the controllers in each carriage for display. Inspections are conducted by dedicated personnel or carriage attendants. However, because freight train carriages are unattended and frequently assembled, wired transmission to manned carriages is not feasible, so this method is rarely used in freight trains. The latter detects axle temperature by installing infrared probes every 30 km. This method is susceptible to external environmental influences and has difficulty in location, resulting in a low alarm fulfillment rate, a high false alarm rate, and high costs. With the development of wireless sensor network technology, the ability of wireless sensor networks to sense, compute and communicate is used to form a network for real-time monitoring of axle temperature changes in trains during operation, which is of great significance to ensuring the safety of train operation. Wireless sensor network nodes are installed in specific locations in each carriage of the train, and the nodes form a network in a self-organizing manner, which can effectively collect and transmit axle parameters in real time. 1. Introduction to the structure of the system The differences between freight trains and passenger trains are as follows: (1) Passenger train carriages have a fixed power source, so there is no need to consider the power consumption of the system; freight train carriages do not have a power source, so the system applied to freight trains is required to have its own power supply and to minimize the power consumption of the system. (2) Each passenger car has a train attendant who can monitor the detected axle temperature and notify a designated person to handle any abnormalities; freight car carriages are generally unattended, so the collected data needs to be transmitted to the locomotive for processing by technicians. (3) Passenger cars are basically fixed and will not be frequently reassembled, so power lines or communication lines can be connected between passenger cars; freight cars may have different destinations, so they need to be moved frequently. Sometimes a car needs to be dropped, and sometimes a new car needs to be attached, so wired communication between freight cars is impossible. In view of the current problems of axle temperature detection in trains and the characteristics of freight trains that are different from passenger cars, we designed an axle temperature detection system based on a wireless sensor network, the structure of which is shown in Figure 1. The whole system consists of a temperature acquisition unit fixed on each car and a main controller at the head of the train. The temperature acquisition unit is called a wireless sensor network node. After the system is powered on, it works in a self-organizing manner and can add or remove a car at any time without affecting the operation of the whole system (as shown in Figure 1). In addition, the node data is transmitted to the head of the train in a multi-hop manner, and the power of wireless transmission of each node does not need to be too high, so it can effectively save the energy consumption of a single node and balance the energy consumption of the entire network. 2. Composition of Wireless Sensor Network Nodes The structure of the wireless sensor network node is shown in Figure 2. The wireless sensor network node comprises a sensor module, a processor module, a wireless communication module, and a power supply module. The node's function is to collect axle temperature data and transmit it to the central control unit via the wireless sensor network, while also acting as a router connecting other nodes to the central control unit. The sensor module is responsible for data acquisition. In this system, eight temperature sensors are installed on the eight axle boxes of the carriage to detect axle temperature and transmit the temperature data to the processor unit via a 1-Wire bus. The processor unit is the core of the node, primarily responsible for controlling data acquisition, analyzing and calculating the data, and controlling the wireless communication module to transmit the processed data to the central control unit through other network nodes connected to it. Furthermore, the processor module is responsible for processing data from other nodes and forwarding it according to rules. The wireless communication module is responsible for data reception and transmission; this design uses the nRF905 module. The nRF905 is a monolithic wireless radio frequency transceiver chip operating at 433/868/915 MHz, consisting of a frequency synthesizer, receiver demodulator, power amplifier, crystal oscillator, and modulator. It employs high-interference-resistant GFSK modulation, a data rate of 90 kb/s, adjustable transmit power up to +10 dBm, unique carrier detection output (CD), data ready output (DR), and address match output (AM), automatically generates preamble and CRC, and communicates with the microprocessor via an SPI interface, making configuration very convenient. Furthermore, the nRF905 operates within a voltage range of 1.9–3.6 V, with very low current consumption: approximately 11 mA for transmit (-10 dBm output), approximately 12.5 mA for receive, and approximately 2 μA for standby, meeting the system's high-performance, low-power requirements. A power supply module is essential for system operation. Since trucks do not have a stable power supply like buses, this design uses battery power. Based on the low-power design, two AA alkaline batteries can sustain operation for approximately one year. When designing nodes for each carriage, we set the hardware number of the node to the carriage number, thus establishing a one-to-one correspondence between nodes and carriages. This allows the system to automatically identify the carriage upon power-on, greatly facilitating carriage management. 3. System Hardware Introduction The system block diagram is shown in Figure 3. The temperature sensor used is the DS18B20, an integrated digital temperature sensor manufactured by Dallas Semiconductor. Unlike traditional thermistor temperature sensors, it can directly read the measured temperature and can be programmed to provide 9- to 12-bit digital readings according to actual requirements. It can convert temperature values ​​into 9-bit and 12-bit digital values ​​within 93.75 ms and 750 ms respectively. Therefore, the DS18B20 is used. It simplifies the system structure and increases reliability. Furthermore, the chip consumes very little power; it can operate normally by obtaining a small amount of power from the bus (a few μW when idle and a few mW when working) and storing it in the on-chip capacitor, generally without the need for an additional power supply. Most importantly, the sensor outputs digital signals, which can be directly connected to the microcontroller's I/O, making connection very convenient. Because digital signals are transmitted on a single bus, the system has good anti-interference capabilities, high reliability, and long transmission distance. The processor uses the MSP430 series microcontroller, whose most significant feature is its ultra-low power consumption. Operating at 1.8–3.6 V and a 1 MHz clock speed, its current consumption is between 0.1 and 400 μA. RAM consumes only 0.1 μA in power-saving mode and a mere 0.7 μA in wait mode. Power consumption is a bottleneck in wireless sensor networks, and nodes must rely on battery power, making the MSP430F149 the optimal choice for CPU. The MSP430F149 uses a 16-bit RISC architecture, and its rich addressing modes, concise core instructions, high processing speed (8 MHz crystal drive, 125 ns instruction cycle), large number of registers, and on-chip data memory give it powerful processing capabilities. In addition, the MSP430F149 operates in an ambient temperature range of -40 to +85℃, making it adaptable to various harsh environments. The wireless communication module uses the nRF905, whose performance is described in the previous section. This design allows for real-time monitoring of the axle temperature of each freight car, greatly ensuring the safety of railway transportation. 4. System Software Introduction The system software design mainly includes node sending and receiving programs, temperature acquisition programs, and the head unit's sending and receiving programs. The node sending program flowchart is shown in Figure 4: When a substation receives data, it considers the data after the header as valid data. The microcontroller first checks the substation ID number. If the ID number is not its own, it sends it to the next connected node and re-enters the receiving state. Otherwise, it continues to judge the command number to determine the substation's action. If the master station needs data, it judges the sensor number to confirm which sensor data from that station the master station needs. After data analysis, the substation collects and packages the on-site data and sends it to the main station, or activates the parameter adjustment system to adjust parameters, and then re-enters the receiving state. Other procedures are not described in detail here. Furthermore, we have included an interface in the program so that the main controller at the train head can continuously monitor axle temperature while also sending data to the ground receiving station via SMS or other wireless communication methods. This allows the ground station to store the data sent by the train, so that the axle temperature data at the time of an accident can be retrieved. 5. Conclusion This system can be applied to freight railway trains, providing real-time and reliable detection and alarm functions for vehicle axle temperature. In addition, the system has excellent scalability. Installing other sensors at nodes allows for the detection of other parameters in the carriages. For example, installing corresponding sensors inside the carriages can monitor the condition of freight cars, serving as an anti-theft measure; installing pressure sensors at specific locations in the carriages can monitor the load conditions; and installing humidity sensors can detect the humidity inside the carriages, etc. With the advancement of technology, wireless sensor networks will be widely used in railway transportation, enabling the convenient collection, analysis, and storage of various parameters to meet people's ever-increasing demands.