Introduction Modern high-voltage electrical equipment is developing towards high performance, intelligence, and maintenance-free operation. The transformation of electrical equipment from planned maintenance to condition-based maintenance has become an inevitable trend. High-voltage circuit breakers are important and numerous electrical equipment in the power system. The losses caused by circuit breaker failures far exceed the value of the circuit breaker itself [1]. In order to improve the reliability of circuit breaker operation, relevant measures must be taken to prevent problems before they occur. Establishing a condition-based maintenance link for high-voltage circuit breakers can predict fault precursors, overcome the blindness of periodic maintenance, and also explore and accumulate data for manufacturing intelligent maintenance-free circuit breakers. This paper discusses the principles and technologies for realizing online condition monitoring of high-voltage circuit breakers. Circuit breakers are very complex electrical equipment. The deterioration of circuit breakers involves many factors such as heat, electricity, mechanics, and environment. Although the fault monitoring and diagnosis of circuit breakers has been studied for a long time, there is still no perfect and mature circuit breaker life assessment model [2]. Most current online monitoring modes for circuit breakers use low-end single-chip microcomputers as the main controller to form independent monitoring devices. Due to the limitations of single-chip microcomputers in terms of speed, word length, and expansion capability, there are problems such as fewer sampling signals, insufficient data acquisition, lack of recording of motion processes, low analysis accuracy, and simple calculation models, resulting in a lack of persuasiveness in diagnostic conclusions [2]. At the same time, there are also shortcomings such as unfriendly human-computer interaction and poor networking capability. This paper believes that the above problems should be solved from the following three aspects: 1) The reasonable selection of circuit breaker monitoring signals and the selection and installation of advanced sensors are the premise and key to condition diagnosis; 2) Selecting a high-performance CPU as the core of the monitoring device is the basic condition for determining the performance index of the monitoring system; 3) Changing the monitoring device to a monitoring system and adopting a distributed architecture is flexible and convenient, which greatly enhances the intelligence and scalability. The high-voltage circuit breaker online condition monitoring system proposed in this paper adopts a distributed structure. The online monitoring unit uses TI's high-performance digital signal processor DSPVC33 as the main controller, which can record long data waveforms of many key parameters of the circuit breaker, perform comprehensive diagnosis, and has local display and remote network functions. It is suitable for the three major categories of circuit breakers: low-oil or high-oil, sulfur hexafluoride (SF6) and vacuum. 1. Online Monitoring Project of High Voltage Circuit Breakers Feature extraction is the first step in online monitoring. Blindly using too many sensors to collect the state characteristics of circuit breakers is neither realistic nor economical [3]. This paper focuses on the practical application of online monitoring, directly extracts key features, and infers the state and performance of circuit breakers through mathematical analysis. The main projects of online monitoring are: 1.1 Electrical wear of circuit breaker contacts Each opening of the circuit breaker will cause a certain degree of damage to the contacts. The electrical wear of the circuit breaker contacts (also known as electrical life) is an important indicator of the circuit breaker performance. In practice, it has been proven that it is unscientific to consider only the cumulative breaking current and the cumulative number of breaking times for the diagnosis of the electrical life of the circuit breaker contacts. The reasonable method is to assess the amount of contact wear during the breaking current based on the magnitude of the breaking current, and then accumulate the amount of contact burnout during each breaking as the basis for judging the electrical life. For various types of circuit breakers, the allowable number of interruptions under the rated short-circuit breaking current can be determined. Let N be the allowable number of interruptions under the rated short-circuit breaking current, and define the relative electrical life (wear) of the contacts of a brand-new circuit breaker as 100%. Then, the relative wear at each interruption under the rated short-circuit breaking current is 1/N. Based on the N-Ib curve of different circuit breakers, the corresponding allowable number of interruptions Nb for any breaking current Ib can be calculated. The relative electrical wear of a single interruption is then 1/Nb. Thus, the relative electrical wear at any interruption can be calculated, and the relative electrical life of the circuit breaker can also be calculated as L = L1 - å (1/Nb), where L1 is the initial value of the circuit breaker's electrical life, a percentage not greater than 1, and its value is determined by the circuit breaker's operating history. For newly commissioned circuit breakers or those that have undergone major repairs, L1 can be taken as 1. Based on the above methods and experimental results, the electrical life calculation curve used in this paper can be interpolated from the table below. In the table, Ib represents any breaking current, Ie represents the rated short-circuit breaking current, N represents the number of breaking operations under the rated short-circuit breaking current, and Qm represents the relative electrical wear of the contacts at the corresponding breaking current Ib. Table 1. Relative Electrical Wear Formula for Oil-Less Circuit Breakers [IMG=Relative Electrical Wear Formula for Oil-Less Circuit Breakers]/uploadpic/THESIS/2007/12/20071219171937303262O.jpg[/IMG] Table 2. Relative Electrical Wear Formula for SF6 Circuit Breakers [IMG=Relative Electrical Wear Formula for SF6 Circuit Breakers]/uploadpic/THESIS/2007/12/2007121917194226003G.jpg[/IMG] Table 3. Relative Electrical Wear Formula for Vacuum Circuit Breakers [IMG=Relative Electrical Wear Formula for Vacuum Circuit Breakers]/uploadpic/THESIS/2007/12/2007121917201263214X.jpg[/IMG] 1.2 Mechanical Life Monitoring of Circuit Breakers In the fault statistics of high-voltage circuit breakers, the highest probability of fault is operating mechanism failure. According to statistics, 40% to 60% of high-voltage circuit breaker operation accidents are caused by mechanical reasons [3]. 1.2.1 Opening and closing coil current waveform The opening and closing coil current is a key characteristic characterizing the operating performance of the circuit breaker operating mechanism, and the current waveform contains a wealth of information. Figure 1 shows a typical circuit breaker tripping circuit current when breaking short-circuit current. T0 is the time when the opening and closing command arrives, T1 is the time when the iron core starts running, T2 represents the time when the iron core decelerates or stops running after touching the load of the operating mechanism, and T3 can be regarded as the time when the switch auxiliary contact a disconnects the coil circuit. T0-T1 is related to the control power supply and coil resistance. The change of T1-T2 characterizes whether the electromagnet iron core operating mechanism is stuck, tripped, or the mechanical load changes. T2-T3 or T0-T3 can reflect the movement of the operating transmission system. By analyzing the above different characteristic times, the mechanical fault trend of the circuit breaker can be diagnosed, including faults such as failure to open and failure to close. 1.2.2 Time travel characteristic curve of circuit breaker breaking process The time travel characteristic curve of circuit breaker contains a variety of information, including opening and closing time, synchronicity, speed and travel signal. These signals are related to the performance of the arc-extinguishing chamber of the circuit breaker, arc time and recovery of the post-arc medium, and the performance of the opening and closing springs. [IMG=Time travel characteristic curve of circuit breaker breaking process]/uploadpic/THESIS/2007/12/2007121917202176238J.jpg[/IMG] 1.2.3 Vibration waveform of circuit breaker during operation When the circuit breaker is in operation, the movement of each component in the corresponding mechanism will have a vibration pulse on the vibration signal time diagram. For different operations, the order of the vibration pulses of each component is unchanged, and the vibration waveform has good repeatability. Therefore, it is effective to analyze the operating characteristics of the circuit breaker using vibration signals. Vibration analysis method has high sensitivity for diagnosing deformation and lubrication faults of the circuit breaker mechanism. It is also a characteristic manifestation of wear of mechanism and arc-extinguishing chamber, component failure, assembly error, etc. [3>]. During the commissioning of the device, after several routine operations, the vibration waveforms of the circuit breaker can be recorded on-site as a "fingerprint" for diagnostic reference. 2. Sensor Selection for Online Monitoring of High-Voltage Circuit Breakers In summary, the detection of the circuit breaker's operating status can be achieved using three types of sensors: current, travel, and vibration. To avoid affecting the normal operation of the circuit breaker, non-contact detection methods are preferred. The sensors are installed outside the circuit breaker body, have no adjustable parts, do not require on-site adjustment, and are less affected by environmental factors. 2.1 Circuit Breaker Contact Breaking Current Detection: The breaking current of the circuit breaker contacts is an AC current. This paper uses a through-core protection CT. The input signal range is 0-100A, and the peak-to-peak output signal is -5V to +5V. 2.2 Closing and Opening Coil Current Detection: The closing and opening coil current is a DC current. This paper uses a Hall current sensor, installed with an open core, which does not affect the secondary wiring of the circuit breaker. The input signal range is 0-20A, and the output signal is a DC current of 4-20mA. Hall sensors are based on the principle of Hall effect. The Hall effect refers to the phenomenon that if a control current I is passed through the two ends of a semiconductor thin film and a magnetic field with magnetic induction intensity B is applied in the direction perpendicular to the thin film, an electromotive force UH will be generated in the direction perpendicular to the current and the magnetic field. The generated electromotive force is called Hall voltage or Hall potential. When the material and geometric dimensions of the Hall element are determined, the magnitude of the Hall potential is proportional to the control current I and the magnetic induction intensity B, that is, UH=KHIB. For a given Hall element, KH is a constant. Hall sensors have a simple structure, wide frequency response, and can realize non-contact detection. 2.3 Detection of the travel of the moving contact of the circuit breaker The travel of the moving contact is a linear displacement, which is indirectly realized by measuring the angular displacement of the linkage mechanism. This paper uses a rotary photoelectric encoder. The rotary photoelectric encoder is characterized by light weight, small torque, and high reliability, and is widely used in various angular displacement measurement applications. The photoelectric encoder consists of a code disk, a light source, and photoelectric elements [4]. The code disk consists of three tracks. The outermost track is the incremental track, with a considerable number of transparent and opaque sector areas distributed on it. The middle track is the direction-discriminating track, with its sector areas offset by half a sector from the incremental track. The inner track has only one transparent slit. During use, the code disk rotates with the target being measured. Three pairs of light sources and photoelectric elements are arranged symmetrically on both sides of the code disk. The number of pulses output by the photoelectric elements in the outer track reflects the magnitude of the target's rotation angle; the phase difference between the outputs of the outer and middle tracks reflects the target's rotation direction; the slit in the inner track serves as the reference position of the code disk, providing an initial zero-position signal. The number of sector areas on the incremental track determines the encoder's resolution. Light-emitting diodes (LEDs) are generally used as light sources, and silicon photocells or phototransistors are generally used as photoelectric elements. Rotary photoelectric encoders are mounted on the rotating shaft of the circuit breaker linkage mechanism. 2.4 Vibration Signal Detection: Piezoelectric accelerometers are the main type of sensor for vibration measurement. Compared with other types of sensors, it has a series of advantages such as high sensitivity, wide frequency range, large linear dynamic range, small size, and diverse installation methods. Piezoelectric sensors are based on the principle of the piezoelectric effect. The piezoelectric effect refers to the direct piezoelectric effect. When a force is applied to certain dielectric materials along a certain direction, causing them to deform, internal polarization occurs, generating opposite charges on their two surfaces. When the external force is removed, they return to an uncharged state. This phenomenon of mechanical energy being converted into electrical energy is called the direct piezoelectric effect. The main components of a piezoelectric accelerometer include a housing, a piezoelectric element, and a mass block pressed tightly onto the piezoelectric element. Because the mass block generates an inertial force proportional to the acceleration, this force acts on the piezoelectric element, causing the surface of the piezoelectric element to generate a charge proportional to the acceleration due to the direct piezoelectric effect. 2.5 The input device simultaneously detects the relevant node states of the circuit breaker operation and combines these node states with the output signals of the above-mentioned sensors to analyze various mechanical properties. 3. Hardware Structure of High-Voltage Circuit Breaker Online Monitoring System To achieve condition-based maintenance, the high-voltage circuit breaker online monitoring system should possess various functions such as data acquisition, signal analysis, fault diagnosis, and data management, and have a user-friendly human-machine interface. The various functional modules within the system should be able to communicate effectively in real time. The system structure is shown in the figure below. [IMG=Hardware Structure of High-Voltage Circuit Breaker Online Monitoring System]/uploadpic/THESIS/2007/12/2007121917205375595C.jpg[/IMG] The core of the system is a distributed circuit breaker online monitoring device. This device adopts a modular back-plug structure, including a DSP main control board, MMI module, input, switching input, switching output modules, and power supply module. The circuit breaker online monitoring device is installed on-site. It collects sensor output signals, performs online analysis and processing, and displays the equipment status in real time on the LCD panel (MMI); it also sends the data to the host computer (PC) via the CAN bus. The host computer system calls advanced application algorithms to calculate relevant circuit breaker indicators in real time and accumulate historical data, providing historical evaluations. The DSP main control board is an intelligent control module based on the TMS320VC33. Its tasks can be divided into four aspects: information acquisition, processing and analysis, control output, and communication. Due to the large workload of driving the LCD module and displaying data on-site, the display function is separated from the DSP main control board and is independently executed by the MMI module based on the 89C51. The DSP main control board and the MMI module are connected via RS-232, using a simple master-slave serial communication protocol. However, the distance between the DSP main control board and the host computer is relatively large, and the amount of data transmitted is also larger. The host computer needs to perform more rigorous inference based on the received data, so a highly reliable and real-time CAN bus connection is used. The TMS320VC33 DSP chip is a floating-point digital signal processor from the TMS320C3X series launched by Texas Instruments (TI). It is a lower-cost version developed based on the original TMS320C31 floating-point DSP, characterized by high speed, low power consumption, low cost, and ease of development. Because it uses an internal 1.8V and external 3.3V power supply, the power consumption is reduced by about an order of magnitude compared to the previous model. It features a high-performance 32-bit CPU with high-speed floating-point operation capabilities, supporting operating speeds up to 75 MIPS, and 34K×32b (1.1Mb) of on-chip dual static RAM. The on-chip memory can map peripherals, including a serial port, two 32-bit timers, and a DMA; it has a program boot function, allowing programs to be loaded from a low-speed EPROM into the system's high-speed RAM for full-speed execution. The DSP main control board is based on the DSPVC33 and includes memory modules, communication interface modules, watchdog circuits, etc. The DSP's high-speed computing power and operating speed enable high-density sampling of 108 points per cycle of the current signal and high-frequency pulse sampling of the angular displacement travel sensor output. 4. Software Structure Overview The software of the circuit breaker online monitoring device includes DSP main control board software and MMI module software. The DSP main control board software is the core of system control, and the MMI board software depends on the data from the DSP main control board. The DSP main control board program can be divided into four main functional modules: monitoring program, communication program, sampling program, and calculation program. The monitoring program is the main loop, employing a task scheduling mechanism similar to an embedded real-time multi-tasking operating system, and performs self-checks on the device. The sampling program performs high-speed sampling and waveform recording of input quantities, and generates clocks. The calculation program captures and calculates various online monitoring indicators during circuit breaker operation. The communication program encapsulates communication data according to the protocol and sends or receives data. 5. Conclusion Online monitoring of electrical equipment is an extension function of substation automation, and online monitoring of circuit breaker status is a major component. This paper introduces an online circuit breaker monitoring device based on a high-speed floating-point DSP, describing its monitoring functions and the sensing technology used. The device calculates and analyzes the electrical and mechanical life of the high-voltage circuit breaker and evaluates its operating performance by detecting signals such as fault current, opening and closing coil current waveforms, and contact travel. The device has been put into field use, demonstrating stable performance and good results.