Flexible direct-current power supply system based on access of multi-mode power supply device and control method

Abstract

The invention provides a flexible direct-current power supply system based on access of a multi-mode power supply device and a control method, and belongs to the technical field of flexible direct-current power supply, and the system comprises a direct-current traction network module which is used for current interaction of the flexible direct-current power supply system; the multi-mode power supply module is used for providing basic electric energy support, accessing the direct-current traction network module by utilizing an energy storage device, processing regenerative braking energy, feeding excess regenerative braking energy back to a medium-voltage power grid, and performing collaborative management by utilizing a droop control mode; the central cooperative controller module is used for carrying out bidirectional communication connection on the multi-system power supply module, obtaining a droop curve parameter correction value through online optimal power flow analysis, adjusting an offline droop slope pre-stored in a local controller, and carrying out control cooperation on the multi-system power supply module; the problems that control conflicts exist in an existing flexible direct-current power supply system, performance is degraded, unnecessary circulation and extra loss are generated, and global energy efficiency optimization cannot be achieved are solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of flexible DC power supply, and particularly relates to a flexible DC power supply system based on multi-standard power supply device access and a control method. BACKGROUND

[0002] With the deepening of the "double carbon" strategy, China's urban rail transit traction power supply system is undergoing a transformation from traditional single energy supply to multi-source coordination and green and efficient composite system. In order to improve system energy efficiency, recover train regenerative braking energy and enhance the consumption capacity of photovoltaic and other new energy, the industry has connected a variety of advanced power supply and energy recovery devices to the DC traction network. In addition to the basic transformer-uncontrolled rectifier unit (12-pulse or 24-pulse rectifier), the current system generally introduces a variety of multi-standard power supply devices, including super capacitors for transient power support, energy storage batteries for energy time shift, mechanical flywheel energy storage systems, and medium-voltage energy feedback devices (bidirectional converters) that can feed the remaining energy of the DC traction network to the medium-voltage AC ring network, and new energy energy routers for connecting distributed photovoltaic.

[0003] The diversified access of these devices, although theoretically constitutes a flexible DC power supply ecosystem with rich functions, greatly improves the potential flexibility, green power consumption rate and energy saving potential of the system, but in actual engineering applications, it has produced new technical challenges. The core problem is that existing power supply devices are usually provided by different manufacturers, and their control systems are independent and fragmented, with only local optimal control as the goal. Even products from the same manufacturer, basically only consider the effect of a single device as the optimization target, without considering the problem from the system level. The overall lack of top-level architecture design and unified scheduling strategy for collaborative optimization from the global level of the system has led to the following problems: Control conflict and performance degradation: the response characteristics and control targets of each device are inconsistent, and response conflicts may occur under dynamic operating conditions, hindering each other and making the overall control effect of the system worse than expected, or even worse than before the new device was introduced; Unnecessary circulating current and additional loss: differences in output impedance, DC voltage set point or power instruction of different devices can easily cause circulating current between devices in parallel DC traction networks. This circulating current does no work, but it can cause device overload, cable heating and significant additional energy loss, offsetting some or even all of the benefits of energy-saving devices; Unable to achieve global energy efficiency optimization: due to the lack of coordination, the system cannot economically optimally allocate total power according to real-time electricity prices, load demand, new energy generation power and the state of each device (such as SOC), limiting the further improvement of the overall energy efficiency of the system.

[0004] Therefore, it is urgent to break the mode of simply "patching" various new devices together in the existing traction power supply system, and to build a system-level solution capable of deeply integrating and coordinately controlling various heterogeneous power supply devices. SUMMARY

[0005] In view of the above problems in the prior art, the present application provides a flexible DC power supply system based on multi-standard power supply device access and a control method, which solves the problems of control conflict, performance degradation, unnecessary circulating current and additional loss, and inability to achieve global energy efficiency optimization in the existing flexible DC power supply system.

[0006] To achieve the above purposes, the technical scheme adopted by the present application is as follows: on the one hand, a flexible DC power supply system based on multi-standard power supply device access is provided, which comprises a DC traction network module, a multi-standard power supply module, and a central collaborative controller module. The DC traction network module is used for current interaction of the flexible DC power supply system. The multi-standard power supply module is used to provide basic power support and access to the DC traction network module, to process regenerative braking energy, and to convert excess regenerative braking energy on the DC traction network into industrial frequency alternating current to feed back to the medium voltage power grid, and to collaboratively manage by using droop control. The central collaborative controller module is used for bidirectional communication connection with the multi-standard power supply module, online optimal power flow analysis of the flexible DC power supply system, obtaining of droop curve parameter correction values and transmission to the multi-standard power supply module, adjustment of pre-stored offline droop slope, control collaboration of the multi-standard power supply module, and completion of the flexible DC power supply.

[0007] The present application has the following advantages: by setting the multi-standard power supply module and the central collaborative controller module, a centralized-distributed hybrid system architecture is constructed, global collaborative optimization is achieved, control conflict and performance degradation are solved, the on-site consumption rate of regenerative braking energy and new energy is improved, unnecessary circulating current and additional loss are avoided, global energy efficiency optimization is achieved, the stability and reliability of system operation are enhanced, and good scalability and plug-and-play potential are provided. Further, the multi-standard power supply module comprises: A multi-standard power supply device submodule is used to provide basic power support, access to the DC traction network, processing of regenerative braking energy, and conversion of excess regenerative braking energy on the DC traction network into industrial frequency alternating current to feed back to the medium voltage power grid. A substation-level energy management submodule is used to collaboratively manage the multi-standard power supply device submodule by using droop control.

[0008] Further, the multi-mode power supply device sub-module comprises: a traditional rectifier unit, an energy storage device, an energy feedback device, a new energy access device, and a local controller installed in each device; The traditional rectifier unit, the energy storage device, the energy feedback device, and the new energy access device are connected in parallel to the DC traction network through a power electronic converter interface; The traditional rectifier unit comprises a traction transformer and a diode uncontrollable rectifier, and is used to provide basic power support; The energy storage device is used to absorb, release, and smooth the power of regenerative braking energy by connecting to the DC traction network through a bidirectional DC-DC converter or a bidirectional DC-AC converter; The new energy access device is used to connect the energy routing device through a multi-port converter, connect the new energy power generation equipment to the DC traction network and the auxiliary power supply system, and set an energy storage port to store the obtained new energy in the energy storage device; The energy feedback device adopts a bidirectional AC-DC converter, connects to the medium-voltage ring network through the AC side of the bidirectional AC-DC converter, and connects to the DC traction network through the DC side of the bidirectional AC-DC converter, and is used to invert the excess regenerative braking energy on the DC traction network to obtain a power frequency alternating current and feedback to the medium-voltage network; The local controller is installed in each device and is used to pre-store an offline droop slope to control each device; The sub-module of the power management of the substation comprises: A traction working condition optimization unit is used to optimize and distribute the energy flow between the traction energy and the multi-mode power supply device by obtaining the medium-voltage ring network voltage, the energy storage device state, the new energy generation amount, and the train load demand information; A braking working condition optimization unit is used to control the energy storage device to store the train regenerative braking energy, and in combination with the traction working condition optimization unit, to distribute the energy of the flexible DC power supply system; An information acquisition unit is used to acquire the 10kV or 35kV voltage of the station and the adjacent two stations, the output voltage and current of the traditional rectifier unit, the state of charge, the terminal voltage, the output current of the energy storage device and the energy feedback device of the station and the adjacent two stations, and the photovoltaic power generation power, and to receive and execute the optimization instructions issued by the central collaborative control layer; A control output unit is used to generate the start voltage threshold and droop control slope parameters of the energy storage device and the energy feedback device, and the dynamic energy distribution proportion setting value facing the multi-port converter, to generate a scheduling decision, and to collaboratively manage the local controller; The droop control algorithm unit is used for providing key calculation support for energy distribution and scheduling decision by using the no-load voltage accurate identification algorithm, the short-term load prediction algorithm and the off-line optimal power flow method.

[0009] The beneficial effect of the further scheme is that the rectifier equipment and the regenerative energy utilization equipment are regarded as a unified whole by the integrated layered collaborative control of the multi-standard power supply device, and specific division of labor is provided, collaborative control is realized, and control conflicts between devices are avoided.

[0010] Further, the key calculation support for energy distribution and scheduling decision is specifically: Based on the off-line optimal power flow method, the off-line optimal droop slope of the traction substation connected with all installed energy storage devices, energy feedback devices and new energy access devices is calculated to obtain the pre-stored off-line droop slope. The no-load voltage is identified by using the online no-load voltage identification algorithm, the maximum real-time no-load voltage in the interval is set as the droop starting voltage, the minimum adjustment interval is set, the maximum real-time no-load voltage is adjusted in response to the change of the no-load voltage, and the adjusted maximum real-time no-load voltage is obtained. The predicted real-time traction load is obtained by using the short-term load prediction algorithm, the maximum real-time no-load voltage is taken as the initial voltage based on the predicted real-time traction load, the initial voltage is reduced in response to the train load being in the braking state and not being charged, and the initial voltage is increased in response to the train load being in the traction state and not being discharged based on the matching of the charging and discharging and the train load. The energy distribution is performed by using the priority of the traction port, the 400V port and the medium-voltage ring network port in response to the train load being in the traction state, and the energy distribution is performed by using the priority of the 400V port, the medium-voltage ring network port and the energy storage port in response to the train load being in the braking state.

[0011] The beneficial effect of the further scheme is that the energy distribution and scheduling decision are realized by the droop control algorithm, the energy scheduling is performed in the DC traction network, the regenerative braking energy and the photovoltaic power generation energy are first used for the traction of the adjacent train or the load of the station, the feedback power grid or waste is replaced, the on-site consumption mode is used, the energy loss in multiple conversions and transmissions is maximally reduced, the energy saving benefit is further improved, and the impact on the upper power grid is reduced.

[0012] Further, the central collaborative controller module comprises a data acquisition and state perception submodule, a target constraint submodule, a power distribution submodule and an instruction and adjustment submodule. The data acquisition and state perception submodule is configured to utilize a preset industrial communication network to perform bidirectional communication connection on the local controller in the multi-standard power supply module, and to acquire flexible DC power supply system data; The target constraint submodule is configured to set an optimization target and a constraint condition. The optimization target includes minimum upper 110kV substation output, maximum local PV consumption rate, and maximum charge-discharge life of energy storage devices. The constraint condition includes stabilizing the DC bus voltage in a preset rated allowable range, maintaining the state of charge of each energy storage device in a preset safety interval, and controlling the output power of each device to be less than a preset maximum allowable power. The power distribution submodule is configured to utilize a deep reinforcement learning algorithm to perform online optimal power flow analysis on the flexible DC power supply system in each control cycle according to the acquired flexible DC power supply system data, in combination with the optimization target and the constraint condition, to generate droop curve parameter correction values of each substation-level controller. The instruction and adjustment submodule is configured to utilize a unified preset industrial communication network to transmit the droop curve parameter correction values and control instructions of each substation-level controller to each local controller, to perform real-time adjustment on the pre-stored offline droop slope of the local controller of the multi-standard power supply device, to control and coordinate the multi-standard power supply module, and to complete the flexible DC power supply.

[0013] Further, the flexible DC power supply system data includes 10kV or 35kV voltage of all stations, rectifier output voltage, rectifier output current, state of charge of energy storage devices, voltage and current of energy storage devices, energy feedback device current, offline train timetable, PV power, no-load voltage prediction value of all stations, load prediction result of all stations, and control instructions of all substation-level energy management submodules.

[0014] The above further scheme has the beneficial effects that the application utilizes a deep reinforcement learning algorithm as a core optimization engine, processes multiple source information in real time, dynamically solves optimal power distribution instructions and droop parameter correction values, and accurately coordinates the operating points of each device, thereby simultaneously achieving multiple goals such as energy saving, economy, and long device life. The central collaborative controller and the local controller of all power supply devices are interconnected through a communication network to form a hybrid system architecture with centralized information management and distributed power execution, which serves as a physical basis for global optimization.

[0015] In another aspect, a flexible DC power supply control method based on multi-standard power supply device access is provided, comprising the following steps: S1, connecting the multi-standard power supply module with the DC traction network, and adopting a layered architecture to construct a central collaborative control layer and a substation energy management distribution layer; S2, according to the substation energy management distribution layer, using a droop control method to collaboratively manage the energy flow in the multi-standard power supply module, obtaining a dynamic energy distribution ratio set value and an offline droop slope pre-stored by the local controller; S3, according to the central collaborative control layer, using a deep reinforcement learning algorithm to analyze the online optimal power flow, obtaining the optimal power distribution instruction and the droop parameter correction value, and adjusting the offline droop slope pre-stored by the local controller, to collaboratively control the multi-standard power supply module and complete the flexible DC power supply control.

[0016] Further, the S1 comprises the following steps: S101, connecting the multi-standard power supply module with the DC traction network, and adopting a layered architecture, using the local controller of the multi-standard power supply module, combining the offline optimal power flow method, the no-load voltage accurate identification algorithm and the short-term load prediction algorithm, to construct the substation energy management distribution layer; S102, setting a central collaborative controller module, establishing a bidirectional communication connection between the central collaborative controller and the local controller through a pre-set industrial communication network, and combining deep reinforcement learning to construct the central collaborative control layer.

[0017] Further, the S2 comprises the following steps: S201, according to the substation energy management distribution layer, based on the droop control method, using the online no-load voltage identification algorithm to identify the no-load voltage, setting the droop starting voltage as the maximum real-time no-load voltage in the interval, setting the minimum adjustment interval, in response to the change of the no-load voltage, adjusting the maximum real-time no-load voltage to obtain the adjusted maximum real-time no-load voltage; S202, using the short-term load prediction algorithm to obtain the predicted real-time traction load, based on the predicted real-time traction load, taking the adjusted maximum real-time no-load voltage as the initial voltage, based on the matching of the charging and discharging and the load of the train, in response to the train load being in the braking state and not being charged, reducing the initial voltage; in response to the train load being in the traction state and not being discharged, increasing the initial voltage; S203, the energy router 400V port load, traction load and photovoltaic power generation is identified, in response to the train load is traction state, using the priority of traction port, 400V port and medium voltage ring network port energy distribution, in response to the train load is braking state, using the priority of 400V port, medium voltage ring network port and energy storage port energy distribution, get dynamic energy distribution ratio set value; S204, using dynamic energy distribution ratio set value, by optimal coordination control to train traction working condition and braking condition, collaborative management of energy flow in multi-system power supply device; S205, using off-line optimal power flow method, calculate all installation energy storage device, energy feedback device and energy routing device of traction substation off-line optimal droop slope, get the pre-stored off-line droop slope in the local controller of multi-system power supply device.

[0018] Further, the S3 comprises the following steps: S301, according to the central level collaborative control layer, collect flexible DC power supply system data, and set optimization target and constraint condition; S302, according to the collected flexible DC power supply system data, using deep reinforcement learning algorithm, combined with optimization target and constraint condition, in each control cycle, online optimal power flow analysis is carried out on flexible DC power supply system, and the droop curve parameter correction value of each substation level controller is generated; S303, using a unified preset industrial communication network, the droop curve parameter correction value and control instruction of each substation level controller are transmitted to the local controller of multi-system power supply device, the pre-stored off-line droop slope in the local controller is adjusted in real time, the multi-system power supply device is controlled and coordinated, and the flexible DC power supply is completed.

[0019] The beneficial effects of the above further scheme are: the present application improves the robustness of DC power supply control by constructing hierarchical control architecture, and the central controller processes slow optimization scheduling, and the local controller is responsible for fast closed-loop control, realizes clear division of labor, improves processing efficiency, when a device fails, the central controller can quickly recalculate and distribute the power task of the remaining healthy devices, realizes fault-tolerant operation, ensures that the core function of flexible DC power supply does not interrupt, and improves the reliability of the whole flexible DC power supply. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The system structure diagram of the present application.

[0021] Figure 2 The control method step diagram in the present embodiment.

[0022] Figure 3A hierarchical control architecture diagram in the embodiment. DETAILED DESCRIPTION

[0023] The specific embodiments of the present application are described below to facilitate the understanding of the present application for those skilled in the art, but it should be clear that the present application is not limited to the scope of the specific embodiments, and for those skilled in the art, it is obvious that various changes are within the spirit and scope of the present application defined and determined by the appended claims, and all the inventions utilizing the concept of the present application are within the scope of protection.

[0024] Before the embodiment is described, the following terms are explained: SOC: state of charge; CAN bus: controller area network bus; Modbus: bus protocol.

[0025] Embodiment 1 In the embodiment, in order to break through the mode of simply assembling various new devices together in the existing traction power supply system, a system-level solution capable of deeply integrating and coordinately controlling various heterogeneous power supply devices is constructed, and a new flexible DC power supply system based on multi-standard power supply device access and a control method are proposed, which aims to solve the coordination problem between devices from the height of system top-level design, eliminate circulating current, and maximize the energy-saving potential and economic operation benefit of the multi-element power supply system.

[0026] As shown in Figure 1 The present application provides a flexible DC power supply system based on multi-standard power supply device access, comprising: a DC traction network module, a multi-standard power supply module, and a central coordination controller module. The DC traction network module is used for current interaction of the flexible DC power supply system.

[0027] In the embodiment, the DC traction network adopts 1500V or 750V, and is connected with the train and the multi-standard power supply device, respectively.

[0028] The multi-standard power supply module is used for providing basic power support, and using the energy storage device to access the DC traction network module to process the regenerative braking energy, and inversely converting the excess regenerative braking energy on the DC traction network into power frequency alternating current to feed back to the medium voltage power grid, and using the droop control mode to coordinately manage the local controller, comprising: a multi-standard power supply device submodule and a substation-level energy management submodule. The multi-standard power supply device submodule is used for providing basic power support, and accessing the DC traction network to process the regenerative braking energy, and inversely converting the excess regenerative braking energy on the DC traction network into power frequency alternating current to feed back to the medium voltage power grid. The multi-mode power supply device sub-module comprises a traditional rectifier unit, an energy storage device, an energy feedback device and a new energy access device; The traditional rectifier unit, the energy storage device, the energy feedback device and the new energy access device are connected in parallel to the DC traction network through a power electronic converter interface; The traditional rectifier unit comprises a traction transformer and a diode uncontrollable rectifier, and is used for providing basic power support; The energy storage device is used for absorbing, releasing and power smoothing of regenerative braking energy by connecting to the DC traction network through a bidirectional DC-DC converter or a bidirectional DC-AC converter; The new energy access device is used for connecting a new energy power generation equipment to the DC traction network and an auxiliary power supply system through a multi-port converter, and setting an energy storage port to store the obtained new energy in the energy storage device; The energy feedback device adopts a bidirectional AC-DC converter, is connected to a medium voltage loop network through an AC side of the bidirectional AC-DC converter, and is connected to the DC traction network through a DC side of the bidirectional AC-DC converter, and is used for converting the excess regenerative braking energy on the DC traction network into an industrial frequency alternating current and feeding back to the medium voltage network.

[0029] In the embodiment, the multi-mode power supply devices are connected in parallel to the DC traction network through a power electronic converter interface, and comprise: The traditional rectifier unit is composed of a traction transformer and a diode uncontrollable rectifier, and is used for providing basic power support for the system; The energy storage device comprises a super capacitor, a storage battery or a flywheel energy storage system. They are connected to the DC traction network through a bidirectional DC-DC converter or a bidirectional DC-AC converter, and are used for absorbing, releasing and power smoothing of regenerative braking energy; The energy feedback device is a medium voltage energy feedback device, adopts a bidirectional AC-DC converter, is connected to a medium voltage (10kV or 35kV) loop network through an AC side of the bidirectional AC-DC converter, and is connected to the DC traction network through a DC side of the bidirectional AC-DC converter, and is used for converting the excess regenerative braking energy on the DC traction network into an industrial frequency alternating current and feeding back to the medium voltage network; The new energy access device is a photovoltaic energy router, is connected to the DC traction network and a 400V auxiliary power supply system through a multi-port converter, and reserves an energy storage port to store the obtained new energy in the energy storage device.

[0030] The sub-module of the energy management of the substation comprises a traction working condition optimization unit, a braking working condition optimization unit, an information acquisition unit, a control output unit and a droop control algorithm unit; The traction working condition optimization unit is used for optimizing and distributing energy flow between traction energy and multi-standard power supply devices by acquiring medium-voltage looped network voltage, energy storage device state, new energy generation amount and train load demand information. The brake working condition optimization unit is used for regulating and controlling the energy storage device, storing train regenerative braking energy, and combining the traction working condition optimization unit to perform energy distribution for the flexible DC power supply system.

[0031] In the embodiment, at the traction substation level, a substation-level energy management submodule is arranged to realize efficient local energy coordination and distribution control. The substation-level energy management submodule is used for cooperatively managing the multi-standard power supply device submodule by using a droop control mode; the station rectifier equipment and various regenerative energy utilization equipment are regarded as a unified whole, and a unified droop control mode is used to cooperatively manage energy flow in view of the problem that different power supply device external characteristics are inconsistent, so as to provide a clear and controllable execution interface for the global optimization instruction of the upper-layer central cooperative control, and specifically includes: The traction working condition optimization unit and the brake working condition optimization unit are embodied through two typical scenarios. Traction working condition and brake working condition optimization of the substation with energy storage devices: Traction working condition: real-time acquisition of medium-voltage looped network voltage, energy storage device state and train load demand and the like, dynamic optimization and control of the distribution ratio of traction energy between the rectifier port and the energy storage port, preferential use of energy storage equipment discharge under the premise of ensuring power supply reliability, and reduction of rectifier set energy consumption.

[0032] Brake working condition: intelligent regulation and control of the energy storage device based on real-time information acquisition, maximization of absorption of train regenerative braking energy, and storage of the remaining energy as much as possible to reduce energy waste; Traction working condition and brake working condition optimization of the substation with energy routers: Traction working condition: comprehensive multi-source information such as medium-voltage looped network voltage, energy storage state, photovoltaic output and train load, optimization and distribution of energy flow from photovoltaic to traction, rectifier to traction, energy storage to traction and photovoltaic to other loads, and control target is to ensure high photovoltaic consumption rate, reduce rectifier traction energy consumption, and maintain high efficiency of the overall system operation; Brake working condition: the remaining energy generated by train braking can be dynamically distributed among energy storage, 400V system consumption and feedback to the medium-voltage looped network, and optimal coordinated control of the three energy paths is realized through real-time information monitoring and logical judgment. The traction working condition optimization unit and the brake working condition optimization unit are used for energy distribution for the flexible DC power supply system.

[0033] The information collection unit is configured to collect 10kV or 35kV voltage of the station and two adjacent stations, output voltage and current of a traditional rectifier unit, state of charge, terminal voltage, output current and photovoltaic power generation of energy storage devices and energy feedback devices of the station and two adjacent stations, receive and execute optimization instructions issued by a central collaborative control layer. The control output unit is configured to generate starting voltage threshold and droop control slope parameters of the energy storage devices and the energy feedback devices, generate a dynamic energy distribution proportion setting value for the multi-port converter, and generate a scheduling decision to collaboratively manage the local controllers. The droop control algorithm unit is configured to use an open-circuit voltage accurate identification algorithm, a short-term load prediction algorithm and an offline optimal power flow method to provide key calculation support for energy distribution and scheduling decisions.

[0034] In the embodiment, the data that the information collection unit needs to obtain in real time includes 10kV or 35kV voltage of the station and two adjacent stations, output voltage and current of a rectifier unit, state of charge, terminal voltage, output current and photovoltaic power generation of energy storage devices and energy feedback devices of the station and two adjacent stations, receiving and executing optimization instructions issued by a central collaborative control layer, output voltage and current of the rectifier unit, SOC, terminal voltage, output current and photovoltaic power generation of energy storage devices and energy feedback devices of the station and two adjacent stations are all used for load prediction of each port, and 10kV or 35kV voltage of two adjacent stations is used for open-circuit voltage identification. The control output unit mainly generates two types of instructions, namely starting voltage threshold and droop control slope parameters of the energy storage devices and the energy feedback devices, and a dynamic energy distribution proportion setting value for each port. The droop control algorithm unit is configured to provide core algorithm support, rely on an open-circuit voltage accurate identification algorithm, a short-term load prediction algorithm and an offline optimal power flow method, and provide key calculation basis for the above-mentioned energy distribution and scheduling decisions. Based on the offline optimal power flow method, the offline optimal droop slope of a traction substation connected with all energy routing devices installed with energy storage devices, energy feedback devices and new energy access devices is calculated. Then, through an online open-circuit voltage identification algorithm, real-time open-circuit voltage is identified, the starting voltage of the droop is set as the maximum real-time open-circuit voltage in the interval, and a minimum adjustment interval is set to adjust when the open-circuit voltage changes. On the one hand, the short-term load prediction algorithm is used to obtain the predicted real-time traction load, the real-time open-circuit voltage is used as the initial value based on the predicted real-time traction load, the starting voltage of the droop is fine-tuned based on the matching of charging and discharging and the load of the train (when the load of the train is in the braking state but not charged, the initial voltage is reduced; when the load of the train is in the traction state but not discharged, the initial voltage is increased). On the other hand, the load of the energy router 400V port and the power generation of the photovoltaic are identified, and when traction, the power distribution is performed in the priority of the traction port-400V port-medium voltage ring network port, and when braking, the power distribution is performed in the priority of 400V-medium voltage ring network-energy storage.

[0035] In this embodiment, the no-load voltage accurate identification algorithm, the short-term load prediction algorithm and the offline optimal power flow method are all existing methods; The no-load voltage accurate identification algorithm adopts the strategy of fusion of formula method and extrapolation method, and the real-time no-load voltage is calculated by formula method, and the value calculated by extrapolation method is used to correct the parameters of formula method. The short-term load prediction algorithm: based on the sampling device of the substation itself and the sampling device of the energy storage system installed in the substation, the ground information, i.e. the original features, can be sampled, and the original features that can be collected include substation voltage, current, no-load voltage, etc. Based on the clustering algorithm, the collected original features are clustered, the original features are divided into clusters and the corresponding cluster centers are obtained, from which the distribution of the original data, i.e. which cluster and the corresponding Euclidean distance from the cluster center, is obtained, which is used as the result of the first feature extraction; the first feature extraction result and the original features are combined as the final features input into the deep neural network, and the overall power of the train in the power supply area obtained offline at the same time scale is used as the label, the deep neural network is trained to obtain the relationship between the features and the label, i.e. the internal relationship between the ground features and the train working condition, and the short-term load prediction is completed. The offline optimal power flow method: by establishing a unified "energy storage-substation" integrated equivalent model, the system energy consumption optimization problem is converted into the solution of the virtual control parameters (reference voltage and equivalent internal resistance) of each site, and a double-layer strategy combining genetic algorithm and traditional nonlinear optimization is adopted to efficiently and reliably find the global optimal solution, so as to minimize the total energy consumption of the entire traction power supply system.

[0036] The central cooperative controller module is used for bidirectional communication connection of the local controllers in the multi-standard power supply module, online optimal power flow analysis of the flexible DC power supply system, obtaining of droop curve parameter correction values and transmission to the multi-standard power supply module, adjustment of the pre-stored offline droop slope of the local controller, control cooperation of the multi-standard power supply module, and completion of the flexible DC power supply, including a data acquisition and state perception submodule, a target constraint submodule, a power distribution submodule, and an instruction and adjustment submodule. The data acquisition and state perception submodule is used for bidirectional communication connection of the local controllers in the multi-standard power supply module by using a preset industrial communication network, and acquisition of flexible DC power supply system data. The target constraint submodule is used for setting optimization targets and constraint conditions. The optimization target includes: minimum output of the upper 110kV main substation, maximum local consumption rate of photovoltaic, and maximum charge-discharge life of energy storage device; The constraint conditions include: stabilizing the DC bus voltage in a preset rated allowable range, maintaining the state of charge of each energy storage device in a preset safety interval, and controlling the output power of each device to be less than a preset maximum allowable power.

[0037] In this embodiment, the central cooperative controller module uses the central cooperative controller to analyze the online optimal power flow of the system based on the deep reinforcement learning algorithm, and the corrected values of the droop curve parameters obtained by the analysis are transmitted to each traction substation to adjust the pre-stored offline droop slope of the local controller. The data acquisition and state perception submodule obtains the 10 / 35kV voltage of all stations, the output voltage of the rectifier, the output current of the rectifier, the SOC of the energy storage device of all energy storage systems, the voltage of the energy storage device, the current of the energy storage / energy feeding device, the offline train timetable, the photovoltaic power, the predicted value of the no-load voltage of all stations, the load prediction result of all stations, and the control instruction of the energy management layer of all substations through an industrial communication network (such as Ethernet, CAN bus, and Modbus, etc.). The obtained data is input into the deep reinforcement learning algorithm, and the real-time electrical quantities and the energy storage SOC constitute the sensory system of the agent to perceive the current system state, directly affecting the decision output of the algorithm; the train timetable, load and photovoltaic prediction information provide the agent with the ability to predict future environmental changes, enabling it to make forward-looking rather than passive decisions; the upper control instruction and the electricity price signal are used to construct the reward function of the algorithm, ensuring that the learned control strategy is highly consistent with our economic and safety goals. The target constraint submodule determines the optimization target: minimum output of the upper 110kV main substation, maximum local consumption rate of photovoltaic, and maximum charge-discharge life of energy storage device; The constraint conditions are determined: the DC bus voltage is stabilized in a preset rated allowable range of 1500V-1900V, the SOC of each energy storage device is maintained in a preset safety interval of 20%-90%, and the output power of each device is less than a preset maximum allowable power.

[0038] The power distribution submodule is used to analyze the online optimal power flow of the flexible DC power supply system in each control period based on the deep reinforcement learning algorithm, combined with the optimization target and the constraint condition, to generate the corrected values of the droop curve parameters of each substation-level controller. The instruction and adjustment submodule is used for transmitting the droop curve parameter correction value and control instruction of each substation level controller to each local controller by using a unified preset industrial communication network, adjusting the pre-stored offline droop slope of the local controller of the multi-standard power supply device in real time, controlling the multi-standard power supply module in coordination, and completing the flexible direct current power supply.

[0039] In the embodiment, the power distribution submodule performs optimization calculation and power distribution, and uses the central controller to run the built-in deep reinforcement learning algorithm based on global real-time and predicted data to perform optimization calculation in each control cycle and generate the droop curve parameter correction value of each substation level controller; The online optimal power flow is specifically realized by modeling and optimization, specifically, the offline optimal power flow is used for modeling, and the deep reinforcement learning algorithm is used for optimization; Typical scenarios are as follows: When a large amount of regenerative energy is generated by train braking, the energy storage system is preferentially instructed to absorb energy; if the energy storage capacity is insufficient or full, the medium-voltage energy feedback device is started to feed the excess energy back to the power grid; When photovoltaic power generation is sufficient and traction load is low, the photovoltaic energy is controlled to be preferentially used for train traction, and the surplus energy is used for charging the energy storage, thereby minimizing the waste of light; By coordinating the external characteristics of the parallel converters, the output voltage-current characteristics are kept consistent, and the condition for generating circulating current is eliminated from the system level; The instruction and adjustment submodule performs instruction issuing and closed-loop adjustment, and issues the control instruction such as the optimization generated droop slope correction value to each local controller through a unified communication protocol, and each multi-standard power supply device adjusts the internal control parameters in real time after receiving the instruction, so as to realize the rapid and accurate distribution and closed-loop execution of power; All multi-standard power supply devices use a unified standardized communication protocol and data format, and the central controller reserves sufficient communication addresses and configuration capacity, so that when a new device is connected, it only needs to complete network registration and parameter reporting, and can be automatically identified and included in collaborative scheduling by the flexible direct current power supply system, and participates in collaborative control, thereby significantly improving the flexibility and scalability of the system.

[0040] In the embodiment, the voltage of each node is adjustable and controllable, and is flexible, and whether it is a station with a medium-voltage energy feedback device, a station with an energy storage device, or a station with an energy router, it has the function of adjusting the voltage; The upper-layer central collaborative controller uniformly manages each device with adjustment function, realizes the adjustable and controllable voltage of each node, optimizes the power flow of the system, uses energy storage and other devices to perform peak clipping and valley filling and new energy absorption, improves the economy and stability of the system, and realizes the flexible direct current power supply.

[0041] In this embodiment, as shown in Figure 1 The multi-standard power supply device installed in the vehicle depot and the multi-standard power supply device installed in the traction substation can constitute a unified flexible DC power supply system based on the multi-standard power supply device access; the photovoltaic and energy storage devices installed in the vehicle depot can be connected to the medium-voltage looped network through the four-quadrant converter and the transformer, and interact with the main line power supply; the energy storage (including flywheel, super capacitor, and battery) and photovoltaic devices installed in the traction substation are connected to the DC traction network through the DC / DC or AC / DC device to interact and regulate energy.

[0042] In this embodiment, the existing DC power supply system is only a simple parallel connection of multiple devices, and lacks a unified command center. The application adds a central collaborative controller to the rail transit flexible DC power supply system. The controller is interconnected with the local controllers of all power supply devices through a communication network, forming a hybrid system architecture with centralized information management and distributed power execution, which serves as the physical basis for global optimization and is a fundamental architecture that differs from the existing decentralized control mode.

[0043] Embodiment 2 As shown in Figure 2 The application provides a flexible DC power supply control method based on multi-standard power supply device access, which is applied to the flexible DC power supply system based on multi-standard power supply device access as described in Embodiment 1, and the implementation method is as follows: S1, connect the multi-standard power supply module using the DC traction network, and adopt a hierarchical architecture to build a central collaborative control layer and a substation-level energy management distribution layer, and the specific steps are as follows: S101, connect the multi-standard power supply module using the DC traction network, and adopt a hierarchical architecture to build a central collaborative control layer and a substation-level energy management distribution layer, and the specific steps are as follows: S102, set up a central collaborative controller module, establish a bidirectional communication connection between the central collaborative controller and the local controller through a pre-set industrial communication network, and build a central collaborative control layer by combining deep reinforcement learning.

[0044] In this embodiment, as shown in Figure 3 The control method adopts a hierarchical architecture, including a central collaborative control layer and a substation-level energy management distribution layer. The upper layer adopts a central collaborative controller, which is responsible for the control and coordination of multiple standard devices in the entire system, and the lower layer adopts a substation-level controller, which is responsible for the real-time control of the devices at the station.

[0045] S2, according to the substation-level energy management distribution layer, the energy flow in the multi-standard power supply module is collaboratively managed by using the droop control method, and a dynamic energy distribution proportion set value and an offline droop slope pre-stored in the local controller are obtained, and the specific steps are as follows: S201, according to the substation-level energy management distribution layer, based on the droop control method, the no-load voltage is identified by using an online no-load voltage identification algorithm, the maximum real-time no-load voltage in the interval is set as the droop starting voltage, the minimum adjustment interval is set, the maximum real-time no-load voltage is adjusted in response to the change of the no-load voltage, and the adjusted maximum real-time no-load voltage is obtained; S202, using a short-term load prediction algorithm, a predicted real-time traction load is obtained, based on the predicted real-time traction load, the adjusted maximum real-time no-load voltage is taken as the initial voltage, based on the matching of charging and discharging and the load of the train, in response to the train load being in the braking state and not being charged, the initial voltage is reduced; in response to the train load being in the traction state and not being discharged, the initial voltage is increased; S203, the energy router 400V port load, the traction load and the photovoltaic power generation power are identified, in response to the train load being in the traction state, the priority of the traction port, the 400V port and the medium voltage ring network port is used for energy distribution, in response to the train load being in the braking state, the priority of the 400V port, the medium voltage ring network port and the energy storage port is used for energy distribution, and a dynamic energy distribution proportion set value is obtained; S204, using the dynamic energy distribution proportion set value, the energy flow in the multi-standard power supply device is collaboratively managed by optimally coordinating the train traction working condition and the braking working condition; S205, using an offline optimal power flow method, the offline optimal droop slope of the traction substation of all installed energy storage devices, energy feedback devices and energy routing devices is calculated, and the offline droop slope pre-stored in the local controller of the multi-standard power supply device is obtained.

[0046] S3, according to the central-level collaborative control layer, the online optimal power flow is analyzed by using a deep reinforcement learning algorithm, the optimal power distribution instruction and the droop parameter correction value are obtained, the offline droop slope pre-stored in the local controller is adjusted, the multi-standard power supply module is controlled collaboratively, and the flexible DC power supply control is completed, and the specific steps are as follows: S301, according to the central-level collaborative control layer, the flexible DC power supply system data is collected, and the optimization target and the constraint condition are set; S302, according to the collected flexible DC power supply system data, using the deep reinforcement learning algorithm, combining the optimization target and the constraint condition, the online optimal power flow of the flexible DC power supply system is analyzed in each control period, and the droop curve parameter correction value of each substation-level controller is generated; S303. Using a unified preset industrial communication network, the droop curve parameter correction values ​​and control commands of each substation-level controller are transmitted to the local controller of the multi-mode power supply device. By adjusting the offline droop slope stored in the local controller in real time, the multi-mode power supply device is controlled and coordinated to complete the flexible DC power supply.

[0047] In this embodiment, addressing the core pain point that the external characteristics and response commands of different power supply devices (rectifiers, energy storage, energy feeders, and photovoltaic routers, etc.) are inherently inconsistent, this invention innovatively proposes a unified integrated droop control method. By assigning appropriate droop slope and voltage reference values ​​to all devices through a central controller, the external characteristics of heterogeneous devices are unified at the control level, laying a methodological foundation for multi-device collaboration. It employs advanced algorithms such as Deep Reinforcement Learning (DRL) as its core optimization engine, enabling it to process multi-source information (load, photovoltaic, electricity price, and SOC information, etc.) in real time, dynamically solve for the optimal power allocation command and droop parameter correction value, and accurately coordinate the operating points of each device, thereby simultaneously achieving multiple goals such as energy saving, economy, and equipment longevity.

[0048] In this embodiment, as Figure 2 The flexible DC power supply control method based on the access of multiple power supply devices provided in the embodiment shown can execute the technical solution shown in the above system embodiment 1. Its implementation principle and beneficial effects are similar, and will not be repeated here.

[0049] In this embodiment of the invention, the flexible DC power supply control method based on multi-standard power supply device access, in order to achieve the principle and beneficial effects of the above-described system embodiment 1, includes hardware structures and / or software modules corresponding to each function. Those skilled in the art should readily recognize that, in conjunction with the illustrative units and algorithm steps described in the embodiments disclosed herein, the present invention can be implemented in hardware and / or a combination of hardware and computer software. Whether a function is executed by hardware or computer software depends on the specific application and design constraints of the technical solution. Different methods can be used to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

Claims

1. A flexible DC power supply system based on multi-standard power supply device access, characterized in that, The application relates to a flexible DC power supply system, which comprises a DC traction network module, a multi-standard power supply module and a central cooperative controller module. The DC traction network module is used for current interaction of the flexible DC power supply system. The multi-standard power supply module is used for providing basic power support and connecting the DC traction network module, processing regenerative braking energy, inversely converting excess regenerative braking energy on the DC traction network into industrial frequency alternating current and feeding back to a medium voltage power grid, and cooperatively managing by using a droop control mode. The central cooperative controller module is used for bidirectional communication connection with the multi-standard power supply module, online optimal power flow analysis of the flexible DC power supply system, transmission of droop curve parameter correction values to the multi-standard power supply module, adjustment of pre-stored offline droop slopes, control cooperation of the multi-standard power supply module and completion of the flexible DC power supply. The multi-standard power supply module comprises a multi-standard power supply device sub-module, a substation level energy management sub-module, a traditional rectifier unit, an energy storage type device, an energy feedback device and a new energy access device.

2. The multi-standard power supply device access based flexible DC power supply system according to claim 1, characterized in that, The traditional rectifier unit, the energy storage type device, the energy feedback device and the new energy access device are connected in parallel to the DC traction network through power electronic converter interfaces. The traditional rectifier unit comprises a traction transformer and a diode uncontrollable rectifier and is used for providing basic power support. The energy storage type device is connected to the DC traction network through a bidirectional DC-DC converter or a bidirectional DC-AC converter and is used for absorbing, releasing and power smoothing processing of regenerative braking energy.

3. The multi-standard power supply device access based flexible DC power supply system of claim 2, wherein, The new energy access device is connected to the DC traction network and an auxiliary power supply system through a multi-port converter and an energy routing device and sets an energy storage port to store obtained new energy into the energy storage type device. The energy feedback device is connected to the medium voltage ring network through an AC side of a bidirectional AC-DC converter and is connected to the DC traction network through a DC side of the bidirectional AC-DC converter and is used for inversely converting excess regenerative braking energy on the DC traction network into industrial frequency alternating current and feeding back to the medium voltage power grid. The local controller is installed in each device and is used for pre-storing offline droop slopes and controlling each device. The substation level energy management sub-module comprises a traction working condition optimization unit and a braking working condition optimization unit. The traction working condition optimization unit is used for optimizing and distributing energy flow between traction energy and the multi-standard power supply device by obtaining medium voltage ring network voltage, energy storage type device state, new energy generation amount and train load demand information. The braking working condition optimization unit is used for regulating and controlling the energy storage type device, storing train regenerative braking energy and combining the traction working condition optimization unit to distribute energy of the flexible DC power supply system. ​ ​ ​ ​ The information acquisition unit is used to collect the 10kV or 35kV voltage of this station and two adjacent stations, the output voltage and current of traditional rectifier units, the state of charge, terminal voltage, output current and photovoltaic power generation of energy storage devices and energy feedback devices of this station and two adjacent stations, and to receive and execute optimization instructions issued by the central collaborative control layer. The control output unit is used to generate the start-up voltage threshold and droop control slope parameters of energy storage devices and energy feedback devices, as well as the dynamic energy allocation ratio setting value for multi-port converters, generate scheduling decisions, and perform collaborative management of the local controller. The droop control algorithm unit is used to provide key computational support for energy allocation and scheduling decisions by utilizing the no-load voltage accurate identification algorithm, short-term load forecasting algorithm, and offline optimal power flow method.

4. The multi-standard power supply device access based flexible DC power supply system of claim 3, wherein, The key computational support provided for energy allocation and scheduling decisions includes: Based on the offline optimal power flow method, the offline optimal droop slope of all traction substations with energy routing devices connected to energy storage devices, energy feedback devices, and new energy access devices is calculated, and the pre-stored offline droop slope is obtained. The online no-load voltage identification algorithm is used to identify the no-load voltage, and the drooping starting voltage is set as the maximum real-time no-load voltage within the interval. The minimum adjustment interval is set, and the maximum real-time no-load voltage is adjusted in response to changes in the no-load voltage to obtain the adjusted maximum real-time no-load voltage. Using a short-term load forecasting algorithm, the predicted real-time traction load is obtained. Based on the predicted real-time traction load, the maximum real-time no-load voltage is used as the initial voltage. Based on the charging and discharging and the matching of the train load, the initial voltage is reduced when the train load is in a braking state and not charging, and the initial voltage is increased when the train load is in a traction state and not discharging. The load on the 400V port of the energy router, the traction load, and the power generation of the photovoltaic system are identified. When the train load is in traction mode, energy is allocated according to the priority of the traction port, the 400V port, and the medium-voltage ring network port. When the train load is in braking mode, energy is allocated according to the priority of the 400V port, the medium-voltage ring network port, and the energy storage port.

5. The multi-standard power supply device access based flexible DC power supply system of claim 1, wherein, The central collaborative controller module includes: a data acquisition and status awareness submodule, a target constraint submodule, a power allocation submodule, and a command and regulation submodule; The data acquisition and status awareness submodule is used to establish a bidirectional communication connection with the local controller in the multi-mode power supply module using a preset industrial communication network, and to acquire data from the flexible DC power supply system. The target constraint submodule is used to set the optimization target and constraint conditions; The optimization objectives include: minimum output of the upstream 110kV main substation, maximum local photovoltaic absorption rate, and maximum charge and discharge life of energy storage devices. The constraints include: stabilizing the DC bus voltage within the preset rated allowable range, maintaining the state of charge of each energy storage device within the preset safe range, and controlling the output power of each device to not exceed the preset maximum allowable power. The power distribution submodule is configured to use a deep reinforcement learning algorithm to analyze online optimal power flow of the flexible DC power supply system in each control cycle according to the collected flexible DC power supply system data, in combination with an optimization target and a constraint condition, to generate a droop curve parameter correction value of each substation-level controller; The instruction and adjustment submodule is configured to use a unified preset industrial communication network to transmit the droop curve parameter correction value and the control instruction of each substation-level controller to each local controller, to control and coordinate the multi-standard power supply module by adjusting the offline droop slope pre-stored in the local controller of the multi-standard power supply device in real time, and to complete the flexible DC power supply.

6. The multi-standard power supply device access based flexible DC power supply system of claim 5, wherein, The flexible DC power supply system data includes 10kV or 35kV voltage of all stations, rectifier output voltage, rectifier output current, state of charge of energy storage devices, voltage and current of energy storage devices, current of energy feedback devices, offline train timetable, photovoltaic power generation power, predicted value of no-load voltage of all stations, predicted result of load of all stations, and control instruction of all substation-level energy management submodules.

7. A method for controlling flexible DC power supply based on multi-standard power supply device access, applied to the flexible DC power supply system based on multi-standard power supply device access according to any one of claims 1-6, characterized in that, The method comprises the following steps: S1, connecting the multi-standard power supply module by using a DC traction network, and using a hierarchical architecture to construct a central-level collaborative control layer and a substation-level energy management distribution layer; S2, using a droop control method to collaboratively manage energy flow in the multi-standard power supply module according to the substation-level energy management distribution layer, to obtain a dynamic energy distribution ratio setting value and an offline droop slope pre-stored in the local controller; S3, using a deep reinforcement learning algorithm to analyze online optimal power flow according to the central-level collaborative control layer, to obtain optimal power distribution instructions and droop parameter correction values, and to adjust the offline droop slope pre-stored in the local controller, to control and coordinate the multi-standard power supply module, and to complete the flexible DC power supply control. 8.The method of claim 7, wherein, The S1 comprises the following steps: S101, connecting the multi-standard power supply module by using a DC traction network, and using a hierarchical architecture to construct a substation-level energy management distribution layer by using a local controller of the multi-standard power supply module in combination with an offline optimal power flow method, an accurate no-load voltage identification algorithm, and a short-term load prediction algorithm; S102, setting a central collaborative controller module, establishing a bidirectional communication connection between the central collaborative controller and the local controller through a preset industrial communication network, and constructing a central-level collaborative control layer in combination with deep reinforcement learning.

9. The method of claim 8, wherein the method further comprises: The S2 comprises the following steps: S201, identifying a no-load voltage by using an online no-load voltage identification algorithm based on a droop control method according to the substation-level energy management distribution layer, setting a maximum real-time no-load voltage in an interval as a droop starting voltage, setting a minimum adjustment interval, adjusting the maximum real-time no-load voltage in response to a change in the no-load voltage, and obtaining an adjusted maximum real-time no-load voltage; S202, using a short-term load prediction algorithm, obtaining a predicted real-time traction load, based on the predicted real-time traction load, taking the adjusted maximum real-time empty voltage as an initial voltage, based on the matching of charging and discharging and the load of the train, in response to the train load being in a braking state and not being charged, reducing the initial voltage; in response to the train load being in a traction state and not being discharged, increasing the initial voltage; S203, identifying the energy router 400V port load, traction load and photovoltaic power generation power, in response to the train load being in a traction state, using the priority of the traction port, 400V port and medium voltage ring network port to perform energy distribution, in response to the train load being in a braking state, using the priority of the 400V port, medium voltage ring network port and energy storage port to perform energy distribution, obtaining a dynamic energy distribution ratio set value; S204, using the dynamic energy distribution ratio set value, through optimal coordinated control of train traction and braking conditions, cooperatively managing energy flow in the multi-standard power supply device; S205, using an offline optimal power flow method, calculating the offline optimal droop slope of the traction substation of all installed energy storage devices, energy feedback devices and energy routing devices, obtaining the pre-stored offline droop slope in the local controller of the multi-standard power supply device.

10. The method of claim 9, wherein the method further comprises: The S3 includes the following steps: S301, according to the central level cooperative control layer, collecting flexible DC power supply system data, and setting optimization targets and constraint conditions; S302, according to the collected flexible DC power supply system data, using a deep reinforcement learning algorithm, combining the optimization target and the constraint condition, performing online optimal power flow analysis on the flexible DC power supply system in each control period, and generating a droop curve parameter correction value of each substation level controller; S303, using a unified preset industrial communication network, transmitting the droop curve parameter correction value and control instruction of each substation level controller to the local controller of the multi-standard power supply device, adjusting the pre-stored offline droop slope in the local controller in real time, cooperatively controlling the multi-standard power supply device, and completing the flexible DC power supply.

Images