Intelligent electrical drive system for lifting and hoisting of wind power installation vessel based on DC bus

Abstract

The present application relates to the technical field of electrical drive and intelligent operation and maintenance of marine equipment and large special ships, in particular to a wind power installation ship lifting and hoisting intelligent electrical drive system based on a DC bus, comprising a DC main bus, a pile leg lifting system, a pile hoisting crane system and a DP positioning system, which are all connected in parallel to the bus. The lifting system adopts 4N+X cross redundancy, and each power supply layer is driven by left and right branch buses to opposite pile legs, and the remaining healthy motor supplies torque when a fault occurs. The crane system adopts 3N multi-bus redundancy, and the core motor is connected to three branch buses, and the remaining buses complete safe winding and unwinding when a fault occurs. The potential and kinetic energy bidirectional recovery module feeds the regenerated electric energy back to the bus. The multi-modal intelligent monitoring module collects current, voltage and vibration signals, extracts multi-branch features and discriminates faults through attention fusion, and supports maintenance without stopping. The present application improves the energy efficiency of redundancy and has predictive maintenance capability.

Description

Technical Field

[0001] This invention relates to the field of electrical drive and intelligent operation and maintenance control technology for marine equipment and large special vessels, and particularly to an intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus. Background Technology

[0002] As offshore wind power expands into deep waters, the power capacity and reliability requirements of the three major systems on wind turbine installation vessels—leg lifting, pile-riding crane, and DP positioning—are increasing. Power allocation directly impacts operational safety. Currently, each of these systems is equipped with independent generator sets and converters, with isolated electrical topologies and separate DC buses, preventing cross-system energy dispatch. During lifting or lowering, regenerated electricity lacks a feedback channel and is dissipated as heat through braking resistors, resulting in waste and a heat dissipation burden, and cannot be absorbed by other systems. In terms of reliability, traditional lifting systems lack a high-redundancy bus architecture and fault reconfiguration capabilities. A single branch failure can cause the corresponding leg to lose strength, and the platform risks overturning due to uneven stress. The crane's core and auxiliary mechanisms share the same bus; a bus failure leads to indiscriminate shutdown, rendering the core mechanism unable to maintain operation, resulting in extremely high safety risks during suspended or emergency lowering of loads. The root cause lies in the fact that the distributed power distribution architecture severs energy interaction and redundancy reconfiguration capabilities between systems, making it difficult to meet the dual requirements of high safety redundancy and efficient energy utilization in deep-sea operations. Summary of the Invention

[0003] In view of this, the present invention proposes an intelligent electric drive system for lifting and hoisting wind turbine installation vessels based on DC bus, in order to solve the technical problems of traditional wind turbine installation vessels where each power system is dispersed and independent, energy cannot be dispatched and recovered across systems, and there is a lack of redundancy backup for single faults, resulting in a high risk of platform overturning or heavy objects falling.

[0004] To achieve the above objectives, this invention designs an intelligent electrical drive system for lifting and hoisting wind turbine installation vessels based on a DC bus, comprising:

[0005] The DC main busbar is where the AC power generated by the ship's main generator sets is rectified to 1000VDC and then fed into the DC main busbar.

[0006] The pile leg lifting system includes four DC power supply layers connected in parallel to the DC main bus and four pile legs distributed on the wind turbine installation vessel. Each pile leg consists of three chords, and each chord is driven by a lifting motor unit, which consists of multiple motors. The four pile legs are named the first pile leg, the second pile leg, the third pile leg, and the fourth pile leg, respectively. The first pile leg and the third pile leg are diagonally distributed to form the first diagonal pile leg group, and the second pile leg and the fourth pile leg are diagonally distributed to form the second diagonal pile leg group.

[0007] Each DC power supply layer is divided into a left branch bus and a right branch bus, and the left branch bus and the right branch bus are connected in parallel to the DC main bus. The left branch bus is connected to the motor of the first diagonal pile leg group through an air circuit breaker and a DC-AC bidirectional inverter, and the right branch bus is connected to the motor of the second diagonal pile leg group through an air circuit breaker and a DC-AC bidirectional inverter, so that the left branch bus and the right branch bus can independently drive the lifting motor group on the first diagonal pile leg group and the second diagonal pile leg group respectively.

[0008] The pile driving crane system includes multiple core mechanisms, auxiliary mechanisms, and three sets of branch buses connected in parallel to the DC main bus. Each core mechanism is driven by three sets of core motors, each core motor consisting of multiple core motors. The three sets of core motors are connected in parallel to the three sets of branch buses. The auxiliary mechanism is driven by one set of auxiliary motors, each auxiliary motor consisting of multiple auxiliary motors. The auxiliary motors are connected to only one set of branch buses. Both the core motors and auxiliary motors are connected to the branch buses through air circuit breakers and DC-AC bidirectional inverters. This ensures that when a core motor on any set of branch buses fails and is isolated, each core mechanism is still driven by the core motor on the remaining branch bus, while the auxiliary mechanism stops and remains in position along with the branch bus that failed.

[0009] The DP positioning system is connected in parallel to the DC main bus.

[0010] The potential and kinetic energy bidirectional recovery module communicates with the DC-AC bidirectional inverter. When the motor or core motor is detected to be in regenerative power generation mode, it controls the corresponding DC-AC bidirectional inverter to switch from frequency conversion drive mode to active rectification mode. This rectifies the three-phase AC power generated by the motor or core motor into DC power and injects it in reverse into the DC main bus. The module also adjusts the output power of the ship's main generator set according to the transient net power of the DC main bus, so that the DP positioning system increases the power consumption of the feedback power on the DC main bus.

[0011] The multimodal intelligent monitoring and diagnosis module includes current sensors, voltage sensors, and vibration sensors respectively installed on the motor, core motor, and auxiliary motor. The multimodal intelligent monitoring and diagnosis module performs time-frequency conversion and multimodal feature fusion processing on the operating signals of the motor and core motor, judges the fault category based on the processing results, and sends the judgment results to the pile leg lifting system or pile crane system to perform non-stop maintenance control.

[0012] Preferably, when a motor fault is detected on the left or right branch busbar, the pile leg lifting system disconnects the faulty motor from the left or right branch busbar via an air circuit breaker to isolate the fault and calculates the total torque of the missing pile leg chord. The calculation formula is:

[0013]

[0014] in, This represents the number of motors that failed due to malfunction. The output torque of these motors at the moment before the failure occurred;

[0015] The pile leg lifting system will reduce the total torque loss. According to the preset torque smoothing compensation coefficient The remaining healthy motors driving the leg chord at the current level are evenly distributed to the remaining healthy motors driving the leg chord at other DC power supply levels, and the torque command of each remaining healthy motor is updated according to the following formula:

[0016]

[0017] in, This is the updated torque command for a single motor. This refers to the original torque command for a single motor. To drive the total number of remaining healthy motors in this diagonal pile leg assembly, Torque smoothing compensation coefficient to prevent instantaneous overshoot.

[0018] Preferably, the multiple core mechanisms of the pile-slewing crane system are the main hoisting winch, the boom luffing mechanism, the slewing mechanism, the sling crane, and the load winch, and the auxiliary mechanism includes the auxiliary hoisting mechanism.

[0019] Preferably, the three sets of branch busbars are connected in parallel to the DC main busbar, the core motor of the core mechanism is connected to the three sets of branch busbars respectively, and the auxiliary motor of the auxiliary mechanism is connected to only one set of branch busbars.

[0020] Preferably, the core motor units of the core mechanism are connected to three sets of branch busbars respectively. When one set of core motor units fails, the other two sets can achieve safe hoisting under heavy load by adjusting the torque of the healthy core motors. Since the auxiliary mechanism is only connected to one set of branch busbars, the auxiliary mechanism stops and maintains its position when the auxiliary motor unit fails.

[0021] Preferably, the potential energy and kinetic energy bidirectional recovery module is used to collect the rotational speed of the motor or core motor in real time. and electromagnetic torque And according to the formula Calculate mechanical power, when mechanical power A negative value indicates that the motor or core motor is in regenerative power generation mode.

[0022] Preferably, the potential energy and kinetic energy bidirectional recovery module adjusts the output power of the ship's main generator set according to the following power balance equation:

[0023]

[0024] in, This refers to the transient net power of the DC main bus. This refers to the power actively fed into the DC main bus after rectification by the ship's main generator sets. This refers to the feedback power when the motor or core motor is in regenerative power generation mode. This refers to the power consumption of the DP positioning system.

[0025] Preferably, the multimodal state intelligent monitoring and diagnosis module uses current sensors, voltage sensors, and vibration sensors to simultaneously collect current signals, voltage signals, and vibration signals during the operation of the motor or core motor, and records the current signals, voltage signals, and vibration signals as current data. Voltage data and vibration data ; Using a sliding window to analyze current data Voltage data and vibration data Sample segmentation is performed, with a sliding window length of 1024 sampling points and a step size of 1024 sampling points, and the windows do not overlap. Then, a short-time Fourier transform is used to perform time-frequency conversion on each sample segment to obtain the corresponding current time-frequency graph. Voltage time-frequency diagram and vibration time-frequency diagram ;

[0026] The current time-frequency graph was analyzed using a multi-branch deep learning network. Voltage time-frequency diagram and vibration time-frequency diagram Feature extraction is performed to obtain the corresponding current features. Voltage characteristics and vibration characteristics ;

[0027] The multi-branch deep learning network includes current feature extraction branches, voltage feature extraction branches, and vibration feature extraction branches, and:

[0028] ;

[0029] ;

[0030] ;

[0031] in, This represents the mapping function implemented by the convolutional neural network in the current feature extraction branch. This represents the mapping function implemented by the densely connected network in the voltage feature extraction branch. This represents the mapping function implemented by the residual neural network in the vibration feature extraction branch.

[0032] Preferably, the multimodal state intelligent monitoring and diagnosis module uses a cross-modal attention mechanism to analyze current characteristics. Voltage characteristics and vibration characteristics To merge, where:

[0033] Utilizing current characteristics Voltage features as a query vector and vibration characteristics Attention is calculated using the key vector and value vector to obtain current fusion features. ,and:

[0034] ;

[0035] Utilizing voltage characteristics Current features as a query vector and vibration characteristics Attention is calculated using key and value vectors to obtain voltage fusion features. ,and:

[0036] ;

[0037] Utilizing vibration characteristics Current features as a query vector and voltage characteristics Attention is calculated using key vectors and value vectors to obtain vibration fusion features. ,and:

[0038] ;

[0039] Current fusion characteristics Voltage fusion characteristics and vibration fusion characteristics Perform tensor splicing to obtain fused features ,and ;

[0040] in, Represents the attention function. This represents the tensor concatenation function.

[0041] Preferably, the multimodal state intelligent monitoring and diagnosis module integrates features. Multi-layer fusion features are obtained by performing layer-by-layer enhancement through multi-layer residual convolutional connections. ,and ;

[0042] in, This represents a 3×3 convolution operation;

[0043] Utilizing multi-level fusion features Two fully connected layers are used for classification mapping to determine the fault type of the motor or core motor, and the judgment result is sent to the pile leg lifting system or pile crane system to perform non-stop maintenance control.

[0044] The beneficial effects of this invention are as follows:

[0045] I. The pile leg lifting system of this invention adopts a 4N+X redundant architecture. Each DC power supply layer is divided into a left branch bus and a right branch bus, which respectively drive the first pile leg group (first pile leg and third pile leg) and the second pile leg group (second pile leg and fourth pile leg) which are diagonally distributed. Diagonal drive can effectively prevent the wind power installation platform from overturning in the event of a major accident. When a motor on any branch fails, the air circuit breaker disconnects the faulty motor from the bus (left branch bus or right branch bus). The missing torque is evenly distributed according to a preset smoothing compensation coefficient to the remaining healthy motors driving the pile leg chords of that DC power supply layer and the motors driving the pile leg chords of the other three DC power supply layers, realizing maintenance without stopping the machine.

[0046] II. The pile-slewing crane system of this invention adopts a 3N redundancy architecture. Each core mechanism consists of three core motor sets connected in parallel to three branch busbars, while the auxiliary mechanism is connected by only one auxiliary motor set. When any branch core motor fails, the air circuit breaker disconnects to isolate the faulty core motor. The missing torque is evenly distributed to the healthy core motors of the remaining two branch busbars according to a preset smoothing compensation coefficient, thus completing the safe lowering or retrieval of heavy loads. When an auxiliary motor fails, the system stops and maintains its position.

[0047] Third, this invention employs a 1000VDC common DC main bus, with the pile leg lifting system, pile-swinging crane system, and DP positioning system all directly connected in parallel to this bus. When the pile legs are lowered or the crane hook descends, the motor or core motor operates in regenerative power generation mode. The corresponding DC-AC bidirectional inverter switches from frequency conversion drive mode to active rectification mode, rectifying the three-phase AC power generated by the motor into DC power and injecting it in reverse into the DC main bus, where it is absorbed in real-time by the azimuth thruster of the DP positioning system. This solution eliminates the heat dissipation of traditional braking resistors, achieving internal energy circulation and peak shaving / valley filling throughout the ship.

[0048] IV. This invention utilizes current sensors, voltage sensors, and vibration sensors installed on the motor and core motor to synchronously collect current, voltage, and vibration signals during motor operation. Time-frequency conversion yields the corresponding time-frequency graphs. A three-branch deep learning network, composed of convolutional neural networks, densely connected networks, and residual neural networks, is then used to extract current, voltage, and vibration features respectively. A cross-modal attention mechanism fuses these multimodal features into a fused feature, which is then enhanced by multiple layers of residual convolution. Finally, two fully connected layers perform classification mapping to determine the motor fault type. The determination result is sent to the pile leg lifting system or the pile-spinning crane system, where the corresponding system performs non-stop maintenance control, including shutting down the faulty motor, transferring the load to a healthy motor, isolating the faulty bus, or switching to a redundant bus. Attached Figure Description

[0049] Figure 1 This is an overview diagram of the power distribution network of the present invention;

[0050] Figure 2 This is a schematic diagram showing the connection between the pile leg lifting system of the present invention and the DC main bus.

[0051] Figure 3 This is a schematic diagram showing the connection between the DC power supply layer and the pile leg of the present invention;

[0052] Figure 4 For the present invention Figure 3 Schematic diagram of the connection between the left-center branch busbar and the first and third pile legs;

[0053] Figure 5 For the present invention Figure 3 Schematic diagram of the connection between the right branch busbar and the second and fourth pile legs;

[0054] Figure 6 This is a schematic diagram showing the connection between the pile-spinning crane system of the present invention and the DC main bus.

[0055] Figure 7 For the present invention Figure 6 A schematic diagram showing the connection between the central branch busbar, core mechanism, and auxiliary mechanism;

[0056] Figure 8 For the present invention Figure 6 Schematic diagram of the connection between the branch busbar and the core structure;

[0057] Figure 9 This is a schematic diagram showing the connection between the potential energy and kinetic energy bidirectional recovery module and the DC-AC bidirectional inverter of the present invention;

[0058] Figure 10 This is a schematic diagram of the structure of the multimodal state intelligent monitoring and diagnosis module of the present invention;

[0059] Figure 11This is a schematic diagram of the electrical system state feature extraction network structure of the present invention.

[0060] The following are labeled in the diagram: 1. DC main busbar; 2. Pile leg lifting system; 21. DC power supply layer; 211. Left branch busbar; 212. Right branch busbar; 22. Pile leg; 221. First pile leg; 222. Second pile leg; 223. Third pile leg; 224. Fourth pile leg; 23. Motor; 3. Pile crane system; 31. Core mechanism; 32. Auxiliary mechanism; 33. Branch busbar; 34. Core motor; 35. Auxiliary motor; 4. DC-AC bidirectional inverter; 6. Current sensor; 7. Voltage sensor; 8. Vibration sensor; 9. Air circuit breaker. Detailed Implementation

[0061] like Figures 1-11 As shown, this invention provides an intelligent electrical drive system for lifting and hoisting wind turbine installation vessels based on a DC bus, with a 1000VDC main DC bus 1 running through the entire vessel as the energy center. The three-phase AC power generated by the main generator sets of the entire vessel is rectified to 1000VDC by a unified rectifier and then fed into the DC main bus 1.

[0062] The pile leg lifting system 2, the pile winding crane system 3, and the DP positioning system are all directly connected in parallel to the DC main bus 1, forming a global topology. This architecture breaks away from the traditional distributed power distribution method where each system on a wind turbine installation vessel is independently equipped with generator sets and AC transformers and rectifier cabinets. All three systems directly obtain DC power from the DC main bus 1, significantly reducing the weight of redundant electrical equipment throughout the vessel and providing a physical basis for the direct cross-system scheduling and reuse of the gravitational potential energy of the crane and lifting system.

[0063] The control functions of the pile leg lifting system 2 are executed by its host controller (an industrial PLC integrated in the pile leg lifting control cabinet), without the need for an additional independent control unit. This host controller communicates with the DC-AC bidirectional inverters 4 of each DC power supply layer 21 via a redundant fieldbus (CANopen) to achieve centralized control of fault detection, torque compensation, and level maintenance.

[0064] The pile leg lifting system 2 adopts a 4N+X cross-redundancy control drive. "4N+X cross-redundancy control drive" refers to a basic framework consisting of four DC power supply layers 21 and four pile legs 22, defined as "4N". Here, "4" represents the four independent basic DC power supply layers 21, and "N" represents the total number of motors 23 configured in each pile leg chord lifting motor group. Based on this, a cross-power supply and cross-fault-tolerant control strategy is further adopted, defined as "X". Specifically, this includes parallel connections on the DC main bus 1. The wind turbine installation vessel has four DC power supply layers 21 and four pile legs 22 distributed on it. Each pile leg consists of three chords, and each chord is driven by a lifting motor unit, which consists of multiple motors 23. The four pile legs 22 are named as first pile leg 221, second pile leg 222, third pile leg 223 and fourth pile leg 224, respectively. The first pile leg 221 and the third pile leg 223 are diagonally distributed to form the first diagonal pile leg group, and the second pile leg 222 and the fourth pile leg 224 are diagonally distributed to form the second diagonal pile leg group.

[0065] Each DC power supply layer 21 is divided into a left branch bus 211 and a right branch bus 212, and the left branch bus 211 and the right branch bus 212 are connected in parallel on the DC main bus 1. The left branch bus 211 is connected to the motor 23 of the first diagonal pile leg group (first pile leg 221, third pile leg 223) through an air circuit breaker 9 and a DC-AC bidirectional inverter 4. The right branch bus 212 is connected to the motor 23 of the second diagonal pile leg group (second pile leg 222, fourth pile leg 224) through an air circuit breaker 9 and a DC-AC bidirectional inverter 4, so that the left branch bus 211 and the right branch bus 212 can independently drive the lifting motor groups on the first diagonal pile leg group and the second diagonal pile leg group, respectively.

[0066] The diagonal mapping design allows both the left branch bus 211 and the right branch bus 212 to independently drive the corresponding diagonal leg groups; any DC power supply layer 21 can independently drive the lifting and lowering of the four legs 22, and the two branch buses within the group can independently drive the diagonal leg groups, achieving extremely high safety redundancy.

[0067] When a fault (such as a severe short circuit or power failure) is detected in the motor 23 on the left branch bus 211 or the right branch bus 212, the upper controller of the pile leg lifting system 2 sends a trip command to the air circuit breaker 9 that controls the faulty motor 23, and the air circuit breaker 9 disconnects the faulty motor 23 from the branch bus.

[0068] To prevent the wind turbine installation vessel from capsizing, while isolating the fault, the upper controller of the pile leg lifting system 2 calculates the total missing torque of the diagonal pile leg group. The calculation formula is:

[0069]

[0070] in, This refers to the number of motors 23 that were shut down due to malfunction. The output torque of these motors 23 at the moment before the failure occurred;

[0071] The upper controller of the pile leg lifting system 2 will detect the missing total torque. According to the preset torque smoothing compensation coefficient The remaining healthy motors 23 driving the current pile leg chord are evenly distributed to the other healthy motors 23 in the DC power supply layer 21, and the torque command of each remaining healthy motor 23 is updated according to the following formula:

[0072]

[0073] in, For the updated torque command of a single motor 23, The original torque command is 23 for a single motor. To drive the remaining 23 healthy motors of the diagonal pile leg assembly, To prevent instantaneous overshoot, the torque smoothing compensation coefficient is set to a value of 0.3 to 0.8, preset according to the marine operating condition level: 0.6 to 0.8 for calm sea areas and 0.3 to 0.5 for rough sea areas.

[0074] The leg lifting system 2 also includes a leveling control function. The upper controller collects the lifting linear velocity of the four legs 22 in real time (through feedback from the encoders of each motor 23). When the difference in linear velocity between any two legs 22 is greater than the preset maximum speed difference threshold (usually 2% of the rated speed), the controller actively adjusts the speed of the motor 23 corresponding to each leg 22 to force the wind turbine installation vessel platform to remain level.

[0075] The speed adjustment adopts a PI control algorithm, with the speed difference as input and the speed compensation amount as output. The adjustment rate is limited to no more than 5% of the rated speed per second, ensuring that the platform level is maintained within ±0.5° throughout the entire lifting and lowering process.

[0076] The control functions of the pile driving crane system 3 are executed by its group control PLC (a redundant controller installed in the crane's electrical room), eliminating the need for an additional independent control unit. This group control PLC communicates with the DC-AC bidirectional inverters 4 and electric load switches 5 on the three branch busbars 33 via industrial Ethernet to achieve logical control for busbar status monitoring, fault isolation, and power reconfiguration.

[0077] The pile-spinning crane system 3 adopts a 3N DC bus group redundant drive scheme. The "3N DC bus group redundant drive scheme" refers to the construction of a triple redundant power supply architecture with three sets of branch buses 33 connected in parallel on the DC main bus 1, defined as "3N". Here, "3" represents three sets of independent basic branch buses 33, and "N" represents the total number of core motors 34 configured in each core mechanism 31. Specifically, it includes multiple core mechanisms 31, auxiliary mechanisms 32, and three sets of branch buses 33 connected in parallel on the DC main bus 1.

[0078] The multiple core mechanisms 31 are the main hoisting winch, the boom luffing mechanism, the slewing mechanism, the sling hoist, and the load winch. Each core mechanism 31 is driven by three sets of core motors, which consist of multiple core motors 34. The three sets of core motors are connected in parallel to three sets of branch busbars 33. This design ensures that any set of branch busbars 33 can independently drive each core mechanism 31, achieving 3N-level safety redundancy.

[0079] The auxiliary mechanism 32 includes an auxiliary lifting mechanism, which is driven by only one set of auxiliary motors and connected to one set of branch busbars 33. The auxiliary motor set consists of multiple auxiliary motors 35, and the core motor 34 and the auxiliary motors 35 are connected to the branch busbars 33 through the air circuit breaker 9 and the DC-AC bidirectional inverter 4.

[0080] Three sets of branch busbars 33 are connected in parallel to the DC main busbar 1. When the core motor 34 on any set of branch busbars 33 fails, the air circuit breaker 9 corresponding to the core motor 34 is disconnected, isolating the faulty core motor 34. The missing torque is calculated by the total torque calculation formula, and the missing torque is compensated by the core motors 34 on the remaining two sets of branch busbars 33, thereby completing the safe lowering and recovery of heavy load. Since the auxiliary mechanism is only connected to one set of branch busbars 33, when the auxiliary motor 35 fails, the auxiliary mechanism stops and stays in position.

[0081] The DP positioning system is connected in parallel to DC main bus 1;

[0082] The potential energy and kinetic energy bidirectional recovery module is connected to the DC-AC bidirectional inverter 4 for communication. When the motor 23 or the core motor 34 is detected to be in regenerative power generation mode, the corresponding DC-AC bidirectional inverter 4 is controlled to switch from frequency conversion drive mode to active rectification mode. This is to rectify the three-phase AC power generated by the motor 23 or the core motor 34 into DC power and inject it into the DC main bus 1 in reverse. The output power of the ship's main generator set is adjusted according to the transient net power of the DC main bus 1 so that the DP positioning system increases power consumption to absorb the feedback power on the DC main bus 1.

[0083] The potential and kinetic energy bidirectional recovery module collects the rotational speed of motor 23 or core motor 34 in real time. and electromagnetic torque And according to the formula Calculate mechanical power.

[0084] The criteria for determining regenerative power generation operation are: the direction of rotational speed is opposite to the direction of electromagnetic torque, and the mechanical power... ,in The preset threshold is typically 5% of the rated power. This decision logic is applicable to scenarios such as the lowering of the pile legs of a wind turbine installation vessel or the lowering of heavy objects by a crane's main hook.

[0085] When mechanical power When the value is negative, it is determined that motor 23 or core motor 34 is in regenerative power generation mode. The potential energy and kinetic energy bidirectional recovery module controls the corresponding DC-AC bidirectional inverter 4 to switch from frequency conversion drive mode to active rectification (AFE) mode.

[0086] The potential and kinetic energy bidirectional recovery module adjusts the output power of the ship's main generator set according to the following power balance equation:

[0087]

[0088] in, This refers to the transient net power of the DC main bus. This refers to the power actively fed into DC main bus 1 by the ship's main generator sets after rectification. This refers to the feedback power when motor 23 or core motor 34 is in regenerative power generation mode. This refers to the power consumption of the DP positioning system.

[0089] The DP positioning system includes a full-rotation thruster. When feedback energy exists on DC main bus 1, causing the bus voltage to rise, the thruster controller of the DP positioning system prioritizes increasing the thruster speed to increase [power / resources]. Real-time absorption of feedback power. If the thruster has reached its maximum speed but there is still residual feedback power, the residual power is consumed through the braking resistor on the DC main bus 1 to maintain the bus voltage stable within the range of 1000VDC±5%.

[0090] This peak shaving and valley filling mechanism enables renewable energy to be consumed in real time by other systems on the ship without the need for energy storage batteries, thus smoothing out power fluctuations of the main generator and reducing fuel consumption.

[0091] The algorithm of the multimodal intelligent monitoring and diagnosis module runs on an embedded GPU controller (installed in the motor monitoring cabinet and connected to the data acquisition card via a PCIe interface), eliminating the need for an additional independent control unit. This controller communicates in real time with the host controller of the pile leg lifting system 2 and the group control PLC of the pile crane system 3 via an isolated CAN bus, enabling the dissemination of fault diagnosis results and the coordinated execution of maintenance actions.

[0092] The multimodal state intelligent monitoring and diagnosis module includes a current sensor 6, a voltage sensor 7, and a vibration sensor 8 respectively installed on the motor 23, the core motor 34, and the auxiliary motor 35. The multimodal state intelligent monitoring and diagnosis module performs time-frequency conversion and multimodal feature fusion processing on the operating signals of the motor 23 and the core motor 34, judges the fault category based on the processing results, and sends the judgment results to the pile leg lifting system 2 or the pile crane system 3 to perform non-stop maintenance control.

[0093] The multimodal intelligent monitoring and diagnosis module uses current sensor 6, voltage sensor 7, and vibration sensor 8 to synchronously collect current, voltage, and vibration signals during the operation of motor 23 or core motor 34, and records the current, voltage, and vibration signals as current data. Voltage data and vibration data ; Using a sliding window to analyze current data Voltage data and vibration data Sample segmentation is performed; the sliding window has a length of 1024 sampling points and a step size of 1024 sampling points, with no overlap between windows. Then, a short-time Fourier transform is used to perform time-frequency conversion on each sample segment to obtain the corresponding current time-frequency graph. Voltage time-frequency diagram and vibration time-frequency diagram ;

[0094] The multimodal state intelligent monitoring and diagnosis module utilizes a multi-branch deep learning network to analyze the current time-frequency graph. Voltage time-frequency diagram and vibration time-frequency diagram Feature extraction is performed to obtain the corresponding current features. Voltage characteristics and vibration characteristics ;

[0095] The multi-branch deep learning network includes current feature extraction branches, voltage feature extraction branches, and vibration feature extraction branches, and:

[0096] ;

[0097] ;

[0098] ;

[0099] in, This represents the mapping function implemented by the convolutional neural network in the current feature extraction branch. This represents the mapping function implemented by the densely connected network in the voltage feature extraction branch. This represents the mapping function implemented by the residual neural network in the vibration feature extraction branch.

[0100] The multimodal state intelligent monitoring and diagnosis module uses a cross-modal attention mechanism to analyze current characteristics. Voltage characteristics and vibration characteristics To merge, where:

[0101] Utilizing current characteristics Voltage features as a query vector and vibration characteristics Attention is calculated using the key vector and value vector to obtain current fusion features. ,and:

[0102] ;

[0103] Utilizing voltage characteristics Current features as a query vector and vibration characteristics Attention is calculated using key and value vectors to obtain voltage fusion features. ,and:

[0104] ;

[0105] Utilizing vibration characteristics Current features as a query vector and voltage characteristics Attention is calculated using key vectors and value vectors to obtain vibration fusion features. ,and:

[0106] ;

[0107] Current fusion characteristics Voltage fusion characteristics and vibration fusion characteristics Perform tensor splicing to obtain fused features ,and ;

[0108] in, Represents the attention function. This represents the tensor concatenation function.

[0109] The multimodal state intelligent monitoring and diagnosis module integrates features. Multi-layer fusion features are obtained by performing layer-by-layer enhancement through multi-layer residual convolutional connections. ,and: ;

[0110] in, This represents a 3×3 convolution operation.

[0111] Utilizing multi-level fusion features Two fully connected layers are used for classification mapping to determine the fault type of motor 23 or core motor 34, and the judgment result is sent to the pile leg lifting system 2 or the pile crane system 3 to perform non-stop maintenance control.

[0112] After receiving the judgment result, if the fault occurs in the pile leg motor 23, the upper controller of the pile leg lifting system 2 will perform the following non-stop maintenance actions: immediately disconnect the air circuit breaker corresponding to the faulty motor, and at the same time, evenly distribute the load of the faulty motor to the remaining healthy motors driving the pile leg chord at the current level and the remaining healthy motors 23 of the same pile leg chord at other DC power supply levels according to the torque compensation algorithm, and update the torque command of each healthy motor. This allows the leg lifting operation to continue even when the single motor is off.

[0113] After receiving the judgment result, the group control PLC of the pile-spinning crane system 3 will, if the fault occurs in the core motor 34 of the core mechanism 31, disconnect the air circuit breaker of the faulty motor and compensate for the missing torque through the core motors 34 on the remaining two branch busbars according to the compensation algorithm, so that the heavy-load safe lowering or retrieval operation can be carried out without interruption. If the auxiliary motor 35 of the auxiliary mechanism 32 fails, disconnect the air circuit breaker of the corresponding faulty motor and stop the machine to maintain its position.

Claims

1. A smart electrical drive system for lifting and hoisting wind turbine installation vessels based on a DC bus, characterized in that, include: The AC power generated by the main generator set of the whole ship is rectified into 1000VDC and then fed into the DC main bus (1); The pile leg lifting system (2) includes four DC power supply layers (21) connected in parallel on the DC main bus (1) and four pile legs (22) distributed on the wind power installation vessel. Each pile leg consists of three chords, and each chord is driven by a lifting motor unit. The lifting motor unit consists of multiple motors (23). The four pile legs (22) are named the first pile leg (221), the second pile leg (222), the third pile leg (223), and the fourth pile leg (224), respectively. The first pile leg (221) and the third pile leg (223) are diagonally distributed to form the first diagonal pile leg group, and the second pile leg (222) and the fourth pile leg (224) are diagonally distributed to form the second diagonal pile leg group. Each DC power supply layer (21) is divided into a left branch bus (211) and a right branch bus (212), and the left branch bus (211) and the right branch bus (212) are connected in parallel on the DC main bus (1). The left branch bus (211) is connected to the motor (23) of the first diagonal pile leg group through an air circuit breaker (9) and a DC-AC bidirectional inverter (4). The right branch bus (212) is connected to the motor (23) of the second diagonal pile leg group through an air circuit breaker (9) and a DC-AC bidirectional inverter (4), so that the left branch bus (211) and the right branch bus (212) can independently drive the lifting motor groups on the first diagonal pile leg group and the second diagonal pile leg group, respectively. The pile-driving crane system (3) includes multiple core mechanisms (31), auxiliary mechanisms (32), and three sets of branch busbars (33) connected in parallel to the DC main busbar (1). Each core mechanism (31) is driven by three sets of core motor groups, each core motor group consisting of multiple core motors (34). The three sets of core motor groups are connected in parallel to the three sets of branch busbars (33). The auxiliary mechanism (32) is driven by one set of auxiliary motor groups, each auxiliary motor group consisting of multiple auxiliary motors (35). The auxiliary motor unit is connected to only one set of branch bus (33), and the core motor (34) and auxiliary motor (35) are connected to the branch bus (33) through the air circuit breaker (9) and the DC-AC bidirectional inverter (4) so ​​that when the core motor unit on any set of branch bus (33) fails and is isolated, each core mechanism (31) is still driven by the core motor unit corresponding to the remaining branch bus (33), while the auxiliary mechanism (32) stops and stays in place along with the branch bus (33) that failed. The DP positioning system is connected in parallel to the DC main bus (1); The potential energy and kinetic energy bidirectional recovery module is connected to the DC-AC bidirectional inverter (4) for controlling the corresponding DC-AC bidirectional inverter (4) to switch from frequency conversion drive mode to active rectification mode when the motor (23) or core motor (34) is detected to be in regenerative power generation mode, so as to rectify the three-phase AC power generated by the motor (23) or core motor (34) into DC power and inject it into the DC main bus (1) in reverse. The output power of the ship's main generator set is adjusted according to the transient net power of the DC main bus (1) so that the DP positioning system increases the power consumption of the feedback power on the DC main bus (1). The multimodal state intelligent monitoring and diagnosis module includes a current sensor (6), a voltage sensor (7), and a vibration sensor (8) respectively installed on the motor (23), the core motor (34), and the auxiliary motor (35). The multimodal state intelligent monitoring and diagnosis module performs time-frequency conversion and multimodal feature fusion processing on the operating signals of the motor (23) and the core motor (34), judges the fault category based on the processing results, and sends the judgment results to the pile leg lifting system (2) or the pile crane system (3) to perform non-stop maintenance control.

2. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 1, characterized in that, When a fault is detected in the motor (23) on the left branch busbar (211) or the right branch busbar (212), the pile leg lifting system (2) disconnects the faulty motor (23) from the left branch busbar (211) or the right branch busbar (212) via the air circuit breaker (9) to isolate the fault and calculate the total torque of the missing pile leg chord. The calculation formula is: ； in, The number of motors (23) that were shut down due to malfunction; The output torque of these motors (23) at the moment before the failure occurred; The pile leg lifting system (2) will reduce the missing total torque. According to the preset torque smoothing compensation coefficient The remaining healthy motors (23) driving the pile leg chord are evenly distributed to the remaining healthy motors (23) driving the pile leg chord in the current layer and the other DC power supply layers (21), and the torque command of each remaining healthy motor (23) is updated according to the following formula: ； in, For the updated torque command of a single motor (23), For the original command torque of a single motor (23), To drive the total number of remaining healthy motors (23) of the diagonal pile leg group, Torque smoothing compensation coefficient to prevent instantaneous overshoot.

3. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 1, characterized in that, The multiple core mechanisms (31) of the pile crane system (3) are the main hoisting winch, the boom luffing mechanism, the slewing mechanism, the sling crane and the load winch, and the auxiliary mechanism (32) includes the auxiliary hoisting mechanism.

4. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 1, characterized in that, The three branch busbars (33) are connected in parallel to the DC main busbar (1). The core motor (34) of the core mechanism (31) is connected to the three branch busbars (33) respectively, and the auxiliary motor (35) of the auxiliary mechanism (32) is connected to only one branch busbar (33).

5. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 4, characterized in that, The core motor units of the core mechanism are connected to three sets of branch busbars (33). When one set of core motor units fails, the other two sets can achieve safe hoisting under heavy load by adjusting the torque of the healthy core motor (34). Since the auxiliary mechanism is only connected to one set of branch busbars (33), the auxiliary mechanism stops and stays in place when the auxiliary motor unit fails.

6. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 1, characterized in that, The potential energy and kinetic energy bidirectional recovery module is used to collect the rotational speed of the motor (23) or the core motor (34) in real time. and electromagnetic torque And according to the formula Calculate mechanical power, when mechanical power When the value is negative, it is determined that the motor (23) or the core motor (34) is in the regenerative power generation mode.

7. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 6, characterized in that, The potential and kinetic energy bidirectional recovery module adjusts the output power of the ship's main generator set according to the following power balance equation: ； in, This refers to the transient net power of the DC main bus. The power of the ship's main generator sets actively fed into the DC main bus (1) after rectification is the power of the entire ship's main generator sets. This refers to the feedback power when the motor (23) or the core motor (34) is in regenerative power generation mode. This refers to the power consumption of the DP positioning system.

8. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 1, characterized in that, The multimodal state intelligent monitoring and diagnosis module uses a current sensor (6), a voltage sensor (7), and a vibration sensor (8) to synchronously collect current signals, voltage signals, and vibration signals during the operation of the motor (23) or the core motor (34), and records the current signals, voltage signals, and vibration signals as current data. Voltage data and vibration data ; Using a sliding window for current data Voltage data and vibration data The sample is segmented; then, a short-time Fourier transform is used to perform time-frequency conversion on each sample segment to obtain the corresponding current time-frequency graph. Voltage time-frequency diagram and vibration time-frequency diagram ; The current time-frequency graph was analyzed using a multi-branch deep learning network. Voltage time-frequency diagram and vibration time-frequency diagram Feature extraction is performed to obtain the corresponding current features. Voltage characteristics and vibration characteristics ; The multi-branch deep learning network includes a current feature extraction branch, a voltage feature extraction branch, and a vibration feature extraction branch, and: ； ； ； in, This represents the mapping function implemented by the convolutional neural network in the current feature extraction branch. This represents the mapping function implemented by the densely connected network in the voltage feature extraction branch. This represents the mapping function implemented by the residual neural network in the vibration feature extraction branch.

9. The intelligent electrical drive system for lifting and hoisting of wind turbine installation vessels based on DC bus as described in claim 8, characterized in that, The multimodal state intelligent monitoring and diagnosis module uses a cross-modal attention mechanism to analyze current characteristics. Voltage characteristics and vibration characteristics To merge, where: Utilizing current characteristics Voltage features as a query vector and vibration characteristics Attention is calculated using the key vector and value vector to obtain current fusion features. ,and: ； Utilizing voltage characteristics Current features as a query vector and vibration characteristics Attention is calculated using key and value vectors to obtain voltage fusion features. ,and: ； Utilizing vibration characteristics Current features as a query vector and voltage characteristics Attention is calculated using key vectors and value vectors to obtain vibration fusion features. ,and: ； Current fusion characteristics Voltage fusion characteristics and vibration fusion characteristics Perform tensor splicing to obtain fused features ,and ; in, Represents the attention function. This represents the tensor concatenation function.

10. The intelligent electrical drive system for lifting and hoisting of a wind turbine installation vessel based on a DC bus as described in claim 9, characterized in that, The multimodal state intelligent monitoring and diagnosis module integrates features. Multi-layer fusion features are obtained by performing layer-by-layer enhancement through multi-layer residual convolutional connections. ,and ; in, This represents a 3×3 convolution operation; Utilizing multi-level fusion features Two fully connected layers are used for classification mapping to determine the fault category of the motor (23) or core motor (34), and the judgment result is sent to the pile leg lifting system (2) or the pile crane system (3) to perform non-stop maintenance control.

Images