Abstract: This paper introduces the mechanical structure design of a novel intelligent axle counter sensor, as well as the axle counting mechanism design, the sensitive element and its channel status identification algorithm, and how to realize the "fault-led safety" principle in the intelligent axle counter sensor. Keywords: Sensor ; Axle Counter; Algorithm 1 Introduction The reliability of an axle counting system is jointly guaranteed by the reliability of the axle counter sensor and the reliability of the subsequent axle counting processor. Currently, all existing axle counter sensors are non-intelligent, basically providing reliable axle presence/absence information to the subsequent axle counting host, with little preprocessing. Obviously, developing an axle counter sensor with a higher performance-to-price ratio would be extremely beneficial for improving the performance of the axle counting system and reducing the overall system cost. Therefore, we studied a new intelligent axle counting sensor design scheme. 2. Mechanical Structure Design of the Intelligent Axle Counting Sensor Figure 1 shows a schematic diagram of the mechanical structure of the intelligent axle counting sensor. To better adapt the sensor to the actual needs of waterproofing, moisture-proofing, high-temperature protection, and impact and vibration resistance in southern stations, the following structural measures were adopted: ①. Double-layer structure design: The outer layer is a mechanical protective layer, made of thick metal plate, with a relatively large mass to meet the requirements of impact and vibration resistance. ②. The four sides of the outer layer are designed with fine louvers, which can provide shade and rain protection, and also facilitate ventilation and heat dissipation. ③. The fastening frame between the inner and outer layers for fixing the inner layer uses sturdy, non-thermal conductive PVC plastic and rubber pads. The inner box is "suspended" and fastened in the middle. The space between the two boxes is mostly hollow, reducing heat conduction and increasing natural heat dissipation. ④. The inner layer is an all-metal, fully sealed structure. The sensitive element is encapsulated in the outer shell with waterproof materials such as epoxy resin. The connections between the sensitive element and the sensor motherboard, power lines, and communication lines are all waterproof to prevent moisture and rainwater intrusion. ⑤. The bottom is also an overhead structure with drainage holes. 3 Main Algorithm Design The three sensitive elements of all intelligent axle counting sensors are numbered S1, S2, and S3 in sequence (see Figure 2). The axle counting values ​​when the corresponding channel is sampled are (S1), (S2), and (S3). 3.1 Axle Counting Mechanism Design In order to complete the axle counting task in real time, whenever a wheel passes over the sensitive element and generates an electrical signal, the microprocessor of the intelligent axle counting sensor completes the relevant axle counting processing by responding to an external interrupt. The interrupt is set to a rising edge trigger mode. One reason is to prevent the interrupt mechanism from being damaged due to channel failure; the other reason is to avoid the interrupt conflict caused by the three sensitive elements. The system sets a response flag for each channel, which are: marks1 (S1 channel), marks2 (S2 channel), and marks3 (S3 channel). Each time the system initializes or performs a "reset" during section clearing, these three flags are set to 0. When the system responds to an axle counting interruption in a certain channel, the flag for that channel is set to 1. Therefore, after responding to 3 consecutive interruptions, the flag for the faulty channel will definitely be 0. If the distance between the sensing elements is N cm, at a vehicle speed of 200 km/h, its linear velocity is 50 m/s, meaning that one millisecond can travel 5 cm. This means that the time interval between a wheel passing through the three sensing elements of the same intelligent axle counting sensor and causing an axle counting interruption is approximately (N/5) ms. The wheel pair spacing is approximately 70 cm-90 cm, meaning that at a vehicle speed of 200 km/h, the minimum time interval between different wheels passing through the same axle counting sensor and causing an axle counting interruption is 14 ms. When a wheel passes through the three magnetic sensing elements, it causes a corresponding change in the magnetic field near them, generating an induced signal. The closer the three sensors are, the more overlap there is between the three induced signals. At higher vehicle speeds, this will inevitably cause interruption conflicts. The solutions are: ①. Set an appropriate threshold level; ②. Set an appropriate spacing between sensitive elements. (See Figure 3) The wheel diameter is generally 35cm-47cm. Considering the maximum wheel diameter of 47cm, its radius is 23.5cm. The height of the sensitive element is selected to be 1.5cm from the lower edge of the wheel. To avoid interruption conflicts, the spacing N between two sensitive elements can be greater than 2L. As shown in Figure 3, L = (23.52-222)1/2 ≈ 8cm. Considering the relationship between the sensor placement position and the wheel curvature, the spacing N can be taken as 15cm. When the train is traveling at 200KM/h, the time interval between the interruption caused by the three sensitive elements is 3ms. 3.2 Axle Counting Data Processing Design The data buffer related to axle counting processing includes the following flags and memory working units. (1). Working units: ①TBUF (1B) ——- Second counting unit ②CINT (1B) ——- Interrupt counting accumulation unit ③C1 (1B) —————— Axis counting accumulation unit of sensitive element S1 ④C2 (1B) —————— Axis counting accumulation unit of sensitive element S2 ⑤C3 (1B) —————— Axis counting accumulation unit of sensitive element S3 ⑥DERECT1 (1B) - Upward direction (i.e., from S1→S3) marking unit ⑦DERECT2 (1B) - Downward direction (i.e., from S3→S1) marking unit ⑧DERECT3 (1B) - Direction uncertain marking unit During initialization or after clearing this section, all the above flags and working units are cleared to zero. (2) Information format design of intelligent axis counting sensor to axis counting processor The information sent by intelligent axis counting sensor to axis counting processor occupies two bytes: the first byte is the working status, and the second byte is the number of axis counting at that time (0-255). The format of the intelligent axle counter sensor's operating status bytes is as follows: ①.SYS represents the intelligent axle counter sensor's microprocessor self-test status: SYS=0 indicates system normal operation; SYS=1 indicates system malfunction. ②.SES1, SES2, and SES3 represent the operating status of three sensing devices and their channels, respectively: SES1=0 indicates the i-th sensing device and its channel are operating normally; SES1=1 indicates the i-th sensing device and its channel are not operating normally. When this section is occupied, the system is in axle counting mode, and these 3 bits are filled in by the axle counting interrupt handler based on the axle counting status. When this section is cleared, the system enters a self-test loop, and these 3 bits are set according to the system's self-test signal and whether its response is correct. ③.DER represents the direction: DER=0 indicates the vehicle's direction of travel is from S1 to S3; DER=1 indicates the vehicle's direction of travel is from S3 to S1. When this section is cleared and the system enters a self-test loop, this bit is 0. ④. ERR is the axle counting abnormal flag: ERR=0 indicates that the axle counting is normal according to the "majority principle"; ERR=1 indicates that the axle counting result may be problematic. ⑤. T1 and T2 are used to record the time delay (in seconds) since the last interruption. When this section is cleared, after the system enters the self-test loop, the second byte is the check byte (0A5H), which is used to check whether the communication link is normal. (3). The sensitive element and its channel working status identification algorithm uses the flags MARKS1, MARKS2, MARKS3, interrupt count accumulation unit and three channel axle counting accumulation units. It is very simple to judge the working status of sensitive elements and their channels: When the value of CINT is greater than or equal to 3, if MARKS1, MARKS2, and MARKS3 are still 0, then the corresponding channel must be faulty. As verification, the corresponding channel axle counting accumulation unit C1, C2 or C3 must be 0. The entire system identifies and judges the direction of travel in both the sensor and the processor, and is ultimately completed by the axle counting processor. (4). Soft timing mechanism design In order to prevent accidental counting caused by human damage or lightning, the system is equipped with a soft timing mechanism. The specific method is: Whenever an axle counting interruption begins, the microprocessor of the axle counting sensor automatically starts a soft clock timing process. ①. Even if the train speed is 1KM/h (which only occurs when the train starts or is about to stop), it can still run about 28cm per second. At this time, the maximum time delay of the train wheels passing through the two sensitive elements is only about half a second. ②. Human damage or lightning interference is reflected in the axle counting with the characteristics of "isolation" and "large time delay", that is, the time interval between two such events is generally much greater than half a second. Based on the above characteristics and referring to other relevant data, human damage or lightning interference can be judged. After setting up this mechanism, to ensure "fault-oriented safety," regardless of the situation, once the identification result indicates that a train may be entering the section, the section is first treated as "occupied." After the soft timing mechanism and subsequent processing confirm that it is an "accidental" event, the processing is lifted. 4. Conclusion The "fault-oriented safety" principle is reflected in the intelligent axle counting sensor in the following aspects: ①. The most important basis for the reliability of any axle counting system is that the axle count must be accurate. The redundant setting of the sensitive elements and the "majority rule" axle counting algorithm of this system greatly improve the reliability of axle counting. ②. The system has set up a complete self-testing mechanism for μP and channels, and also sets a check code, and includes the self-testing results in the communication data. When the system has a fault that may endanger safety, it can tell the axle counting processor the fault status in real time. This prepares the axle counting processor to control the track relay operation according to the "fault-oriented safety" principle. The innovations of this article are: 1) fully implementing the principle of "fail-oriented safety" and analyzing the hardware composition principle of the intelligent axle counting sensor; 2) a brand-new axle counting mechanism design, namely, 3-sensor signal design; 3) the intelligent axle counting sensor is processed by a soft timing mechanism design, namely, anti-interference design. References: [1] Xue Ruimin, Fu Jun, Kan Huanzhang. CLC (Axle Counting + Loop) Automatic Block System. Railway Communication and Signal, 2000, (09) [2] Peng Yuqiang, Liao Wuzhou. ACE Online Testing and Remote Diagnosis System for Axle Counting Automatic Block System. Railway Communication and Signal, 1999, (09) [3] Liu Shengge. Debugging Method of Axle Counting and Loop Automatic Block System. Railway Communication and Signal, 2003, (11) [4] Liu Zhihong, Wang Gengsheng, Wei Minghua. Intelligent Transportation System (RITS) [J]. Microcomputer Information, 2006, 7-3: 16-19