[b]1 Introduction[/b] The stator of a novel dual-winding induction generator has two sets of windings with the same number of poles. One set is the power winding, with the output end connected to the excitation capacitor and rectifier load; the other set is the control winding, connected to the excitation converter, which can continuously adjust the excitation of the power generation system to keep the output voltage of the power winding constant. The mathematical simulation of the dual-winding induction generator system includes multiple mathematical models such as the generator, power semiconductor, controller, and measuring device. The mathematical simulation of the continuous system ensures that the integral error is within the specified range, and the simulation speed is often improved by dynamically adjusting the simulation step size. In the real-time control of this power generation system, the control program usually runs on embedded devices such as DSPs, completes the control program within a fixed clock cycle, and sends control signals to the actuator—the power semiconductor. This leads to bottleneck problems such as interrupt delay, execution time, hardware interface, and measurement error. The solution to these problems is to connect the dual-winding induction generator, power semiconductor, and sensors into the simulation loop. The simulation system works according to the actual time, which can meet the real-time requirements. This semi-physical simulation form is also called a rapid control prototype. This paper proposes a real-time simulation method for decoupling a discrete-event excitation converter system and a continuous-time dual-winding generator system based on the dSPACE single-board system DS1104 experimental platform. A corresponding experimental model is established, and an experimental study is conducted on a dual-winding induction generator system. [b]2 Control Strategy of the Excitation System[/b] When the speed or load of the dual-winding induction generator changes, the output voltage should be kept constant. This requires adjusting the excitation reactive power of the motor and simultaneously adjusting the active power of the DC-side capacitor of the excitation converter on the control winding side. This paper adopts a stator field-oriented control strategy for the control winding. The excitation control principle diagram of the dual-winding generator system is shown in Figure 1. The entire system control is divided into two closed loops. The output voltage Upd detected at the output terminal of the power winding rectifier bridge is compared with the given reference voltage Updref to obtain an error signal. This error signal is then controlled by a PI regulator to obtain the reference reactive current component setpoint i*sd, which compensates for the excitation reactive power required by the control motor. The error signal obtained by comparing the voltage detection value Ucd of the DC-side capacitor of the control winding excitation converter with the given reference voltage setpoint Ucdref is multiplied by a coefficient -1 after PI regulation to obtain the reference active current component setpoint i*sq, which compensates for the active power required by the control capacitor Cc. The coefficient -1 indicates that the active power flows from the motor to the excitation converter. 3. Real-time System Hardware Design The dSPACE real-time simulation system is a control system development and testing platform based on Matlab/Simulink, seamlessly integrated with Matlab/Simulink. The single-board system DS1104 used in this paper consists of a main processor, auxiliary DSP, interrupt controller, memory, timer, and host interface. The DS1104 controller board uses a PowerPC processor for floating-point operations, with a main frequency of 250 MHz and strong computing power. The DS1104 controller board also integrates an I/O subsystem with TMS320F240DSP as the core to meet special I/O requirements. The dSPACE software system consists of three parts: algorithm development module, real-time operation module, and real-time test and monitoring module. The development of the dual-winding generator control system based on dSPACE includes: (1) LAB/Simulink model establishment and offline simulation. The mathematical model of the dual-winding induction generator simulation object is established using Matlab/Simulink, the control scheme is designed, and the system is simulated offline. (2) I/O access. The modules that need to be downloaded to SPACE are retained in Matlab/Simulink, the original logical connection relationship is replaced by the hardware interface relationship, I/O is configured, and the hardware and software interrupt priorities are set. (3) The code is automatically generated and downloaded using the tools provided by RTW and dSPACE, and the model is converted into a program that can be run by the real-time simulator. (4) dSPACE comprehensive experiment and debugging. Real-time simulation data is obtained using ControlDesk. The powerful real-time code enables the use of Software RTI and the user-friendly experimental software ControlDesk, allowing for the rapid establishment of a real-time control system platform for a dual-winding induction generator. The system hardware connection diagram is shown in Figure 2. The hardware circuit includes a main circuit consisting of an excitation converter, generator, and rectifier load, while the control circuit, centered on the DS1104, includes stator dual-winding voltage and current detection, DC bus voltage detection circuits, and protection circuits. This system design includes 10 A/D sampling circuits, used to sample control-side bus current and voltage, control-side two-phase current (only two phases of the three-phase current are independent), power-side two-phase current and voltage, and power-side bus current and voltage. The required samples are connected to the eight ADC units of the DS1104 via coaxial cables, mainly including: control-side current detection, power-side current detection, control-side voltage detection, and power-side voltage detection. Additionally, a Complex Programmable Logic Device (CPLD) is used to comprehensively process fault signals. This system design contains 16 protection signal inputs. After ANDing, a FAULT signal is generated and input to the main processor of the DS1104 controller board. The main processor generates a hardware interrupt signal, causing the program to stop running in the Matlab software. At the same time, a BRAKE signal is also output to directly turn off the PWM signal on the control platform hardware, realizing dual protection of the experimental platform. After processing, these 16 protection signals output 14 low-level active display signals, causing the corresponding LEDs to light up for alarm. This control scheme is verified on a small prototype fully digital control platform of 1500 rpm, 900 W, which is a self-modified three-phase induction motor. A real-time simulation interface with adjustable PI is designed. The controller system cycle is 80μs, and the width of the digital hysteresis loop is 0.5 A. 4 Experimental Study 4.1 System self-excitation voltage build-up adopts a 105μF self-excitation capacitor for self-excitation voltage build-up. When the DC voltage on the power side reaches 120 V, it switches to field-oriented vector control of the control winding. The self-excitation voltage build-up test waveform under rated speed and no load is shown in Figure 3. Figure 3 shows the waveform of the self-excited voltage build-up test under no-load at rated speed. Because a large filter capacitor is used on the DC output side of the rectifier bridge, the DC voltage on the power side shown in Figure 3 remains flat and smooth after voltage build-up stabilizes, with almost no significant fluctuations. 4.2 Speed ​​Change Process under Rated Load with Excitation Capacitor C=105μF: When the system is under rated load, the speed rapidly increases from 1500 rpm to 2000 rpm. During this process, a voltage-oriented control strategy for the control winding is adopted, and the DC voltage on the power side rises to 9 V, approximately 5.6%, as shown in Figure 4. The waveforms of the control winding voltage and current under rated load are shown in Figure 5. When operating under rated load, the control winding line current lags behind the line voltage by approximately 120° (the phase current lags behind the phase voltage by approximately 90°). Since the voltage and current reference directions are based on motor conventions, the control winding excitation converter acts as a capacitor, providing the generator with the required reactive power. When operating under full load, due to the demagnetizing effect, the excitation capacitor cannot provide the generator with the required reactive power. In this case, control winding compensation is required. Furthermore, as the speed increases, the generator's required reactive power decreases, and the current in the control winding decreases, which conforms to the principle of field weakening as the motor speed increases. [b]5 Conclusion[/b] This paper adopts the real-time simulation method of decoupling the discrete event inverter system and the continuous-time generator system on the dSPACE single-board system DS1104 test platform. The mathematical model of the dual-winding induction generator-inverter-sensor is replaced and directly put into the simulation loop for semi-physical simulation research. The content includes: system voltage build-up, variable speed operation under rated load, etc. The research results show that applying dSPACE to the dual-winding induction generator system is beneficial to shorten the development cycle, reduce development costs, and improve the reliability of system operation.