A power supply energy storage device for a transmission line

Abstract

The present invention provides a new type of power supply energy storage device for transmission lines, which belongs to the field of power system energy storage technology. The device is composed of a hybrid energy storage module, an embedded fault prediction unit, a dynamic balancing controller and a composite heat dissipation shell. The hybrid energy storage module integrates all-vanadium liquid flow batteries, solid-state lithium capacitors and supercapacitors to collaboratively meet the energy and power requirements of the transmission line; the embedded fault prediction unit uses a neural network accelerator to predict faults and respond in advance by monitoring current harmonics and temperature data in real time; the dynamic balancing controller adjusts the working state of the energy storage unit according to the probability of failure to optimize energy management; the composite heat dissipation shell adopts phase change materials and bionic air duct design to achieve efficient dynamic heat dissipation. The present invention adopts the above-mentioned new type of power supply energy storage device for transmission lines, which significantly improves the performance, reliability and safety of the energy storage device, extends its service life, reduces maintenance costs, and is suitable for transmission lines with complex working conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of power system energy storage, and in particular to a novel power supply energy storage device for a transmission line. Background Art

[0002] With the continuous growth of global energy demand and the large-scale integration of renewable energy, the stability and reliability of power transmission lines face severe challenges. Traditional transmission line energy storage devices primarily use a single type of energy storage element, such as lead-acid batteries, lithium-ion batteries, or supercapacitors. However, these single energy storage devices often have certain limitations when facing the complex and changing power transmission environment.

[0003] On the one hand, a single energy storage component cannot meet the energy density, power density, cycle life, and cost-effectiveness requirements of transmission lines. For example, lead-acid batteries, while relatively low in cost, have low energy density and a short cycle life, making them unable to quickly respond to instantaneous power fluctuations in transmission lines. Lithium-ion batteries have high energy density and long cycle life, but their power output capacity is limited and they are prone to performance degradation and safety hazards at high charge and discharge rates. Supercapacitors can provide high power output, but their energy storage capacity is relatively small, making them unable to maintain stable operation of transmission lines for long periods of time.

[0004] On the other hand, most existing energy storage devices lack the ability to monitor their own status and the operational status of transmission lines in real time, allowing for intelligent management. When a transmission line fails, energy storage devices are unable to predict and respond effectively, often leading to a wider range of faults and prolonged recovery times, seriously impacting the reliability and quality of the transmission system.

[0005] Furthermore, some energy storage devices have poor heat dissipation performance. Over extended periods of operation, heat accumulation can easily lead to thermal runaway, further shortening the device's service life and potentially causing safety incidents. Furthermore, existing energy storage devices often have fixed heat dissipation designs, unable to dynamically adjust heat dissipation efficiency based on actual operating conditions. This results in energy waste and suboptimal heat dissipation.

[0006] In existing technologies, such as patent CN203445632U, a new type of power supply energy storage device for transmission lines is installed on the transmission line conductors and consists of a high-performance lithium battery, a DC-DC converter, a super energy storage capacitor, and a switching control circuit. This device switches between different energy supply modes by switching the control circuit to power the monitoring terminal, solving the problem of power supply to the monitoring terminal when the line is out of power during a fault. However, this solution still has some shortcomings:

[0007] 1. Energy storage components are still mainly high-performance lithium batteries and supercapacitors, which do not fully utilize the synergistic advantages of multiple energy storage technologies and cannot simultaneously meet the high energy density and power density requirements of transmission lines.

[0008] 2. The lack of real-time monitoring and intelligent management of the internal status of the energy storage system makes it impossible to predict failures and take measures in advance, affecting the reliability and stability of the system.

[0009] 3. The heat dissipation design of the energy storage device is not mentioned, which may cause heat accumulation problems during long-term operation, affecting the life and safety of the energy storage device. Summary of the Invention

[0010] The purpose of the present invention is to provide a new type of power supply energy storage device for transmission lines, which significantly improves the performance, reliability and safety of the energy storage device, extends its service life, reduces maintenance costs, and is suitable for transmission lines with complex working conditions.

[0011] To achieve the above-mentioned object, the present invention provides a novel power supply energy storage device for a transmission line, comprising: a hybrid energy storage module, an embedded fault prediction unit, a dynamic balancing controller and a composite heat dissipation housing;

[0012] The hybrid energy storage module consists of all-vanadium liquid flow battery units, solid-state lithium capacitor units and supercapacitor units connected in parallel through a bidirectional DC-DC converter. The output ends of each unit are electrically connected through a DC bus.

[0013] The embedded fault prediction unit is connected to the hybrid energy storage module through a data bus, and includes a current harmonic analysis module, a temperature field reconstruction module, and a neural network accelerator. The output of the current harmonic analysis module is connected to the input of the neural network accelerator.

[0014] The dynamic balancing controller is connected to the bidirectional DC-DC converter in the hybrid energy storage module, the active balancing circuit of the solid-state lithium capacitor unit, and the microfluidic pump of the all-vanadium liquid flow battery unit through multiple PWM signal lines. It receives the fault probability value output by the neural network accelerator and generates control instructions.

[0015] The composite heat dissipation shell is wrapped around the outside of the hybrid energy storage module, and the internal bionic air duct is tightly fitted to the surface of the hybrid energy storage module. The heat dissipation efficiency is dynamically adjusted through temperature sensor signal feedback.

[0016] Preferably, the structure of the solid-state lithium capacitor unit includes: the positive electrode material is single crystal LiNi 0.8 Mn 0.1 Co 0.1 O2, specific capacity ≥ 200mAh / g; negative electrode material is silicon carbon nanowire array, diameter ≤ 80nm, loading ≥ 4mg / cm 2 ; The solid electrolyte layer is Li 1.5Al 0.5 Ge 1.5 (PO4)3 film, thickness ≤ 20μm, ionic conductivity ≥ 1mS / cm; the battery cells are connected by copper-aluminum composite foil, and the gaps between adjacent battery cells are filled with aerogel insulation layer with a thermal conductivity coefficient ≤ 0.02W / m·K.

[0017] Preferably, the circuit topology of the hybrid energy storage module includes: a triple interleaved parallel DC-DC converter is used between the all-vanadium liquid flow battery unit and the solid-state lithium capacitor unit, with an efficiency ≥97% and a ripple ≤1%; the supercapacitor unit is directly connected to the DC bus, and high-frequency oscillation is suppressed by distributed micro-inductors with an inductance of 20nH±5%; redundant switching channels are set between each unit, with a switching response time ≤100μs, and the switching trigger signal is transmitted by a dynamic balancing controller through an optocoupler isolation circuit.

[0018] Preferably, the FIR filter bank coefficient update rule of the current harmonic analysis module is as follows:

[0019] Based on historical fault data, the center frequency is dynamically adjusted, and the update formula is:

[0020] ;

[0021] in, represents the new center frequency, represents the old center frequency, represents the learning rate, Indicates the The amplitude of the subharmonics, Indicates the The frequency corresponding to the subharmonic;

[0022] The filter coefficients are updated online via the JTAG interface every 24 hours;

[0023] When updating, a sliding window mechanism is used to select the latest 100,000 groups of sampling points as the update data window.

[0024] Preferably, the current harmonic analysis module collects the line current signal in real time and extracts the characteristic harmonic spectrum data through a programmable FIR filter group, whose center frequency is dynamically configured to 6kHz, 12kHz, and 18kHz, and transmits the extracted characteristic harmonic spectrum data to the neural network accelerator through the SPI interface.

[0025] Preferably, the temperature field reconstruction module collects temperature data through distributed optical fiber sensors and sends the collected temperature data to the neural network accelerator via the I2C bus.

[0026] Preferably, the neural network accelerator has a built-in LSTM-TCN hybrid model to calculate the input harmonic distortion rate, temperature gradient matrix and historical fault time series, and output the fault probability value;

[0027] The weights of the LSTM-TCN hybrid model are trained using the improved whale optimization algorithm:

[0028] Step S1: Initialize the position of the whale group. Each whale represents a set of connection weights and bias parameters of the LSTM-TCN hybrid model.

[0029] Step S2: Define the fitness function, and the calculation formula is:

[0030] ;

[0031] in, represents the fitness value, represents the harmonic mean of the model’s prediction accuracy and recall, represents the time required for the model to calculate the output failure probability value, Indicates the energy consumption during the model operation;

[0032] Step S3: Iteratively update the position of the whale group:

[0033] Calculate the fitness value of each whale;

[0034] Select the best whale based on fitness value;

[0035] Simulate whale encirclement, spiral updating, and prey search behaviors to update the whale group's position;

[0036] Step S4: Introduce gradient information in the iterative process, combine the gradient descent method, calculate the gradient of the loss function with respect to the model parameters, and adjust the parameters along the gradient direction when the whale position is updated;

[0037] Step S5: In each iteration, normalize the model weights;

[0038] Step S6: Repeat steps S2 to S5 until the maximum number of iterations is reached or the convergence condition is met, and output the connection weights and bias parameters of the final optimized LSTM-TCN hybrid model.

[0039] Preferably, the dynamic balancing controller receives the fault probability value output by the neural network accelerator, and when the fault probability value is ≥0.6, sends a three-stage pre-charge instruction to the hybrid energy storage module to control the working state of the supercapacitor unit, the solid-state lithium capacitor unit, and the all-vanadium liquid flow battery unit, wherein the PWM duty cycle of the pre-charge instruction satisfies a linear relationship with the fault probability value:

[0040] ;

[0041] in, Indicates the PWM duty cycle of the precharge instruction, Represents the failure probability value.

[0042] Preferably, the design of the composite heat dissipation housing includes: a phase change material layer of paraffin-expanded graphite-boron nitride composite material, a phase change latent heat ≥ 200 J / g, a thermal conductivity ≥ 18 W / m·K, and a thickness of 2.5 mm ± 0.1 mm; a bionic air duct structure having a shark skin non-smooth surface, a groove depth of 50 μm ± 5 μm, a spacing of 200 μm, and an inclination angle of 55° ± 2°; and a heat dissipation efficiency adjustment logic that satisfies:

[0043] ;

[0044] in, Indicates the heat dissipation efficiency, Indicates the real-time temperature in °C. Represents the base of natural logarithms.

[0045] Therefore, the present invention adopts the above-mentioned new power supply energy storage device for transmission lines, and the beneficial technical effects are as follows:

[0046] (1) Collaborative work of multiple types of energy storage elements: The present invention adopts a hybrid energy storage module consisting of all-vanadium liquid flow battery units, solid-state lithium capacitor units and supercapacitor units, which are connected in parallel through a bidirectional DC-DC converter. This collaborative working mode fully utilizes the advantages of different energy storage technologies. The all-vanadium liquid flow battery units can provide large-scale energy storage to meet the long-term stable power supply requirements of the transmission line; the solid-state lithium capacitor units have high specific capacity and high power density and can charge and discharge quickly under large currents; and the supercapacitor units can instantly provide high power output or absorb energy shocks. Compared with the existing technology, this solution overcomes the limitations of a single energy storage element in terms of energy density and power density, can better cope with the complex and changeable working conditions of the transmission line, and improve the overall performance and reliability of the energy storage system.

[0047] (2) Intelligent fault prediction and dynamic balancing control: The embedded fault prediction unit includes a current harmonic analysis module, a temperature field reconstruction module, and a neural network accelerator, which can collect and analyze line current signals and temperature data in real time. Through the LSTM-TCN hybrid model built into the neural network accelerator, the input harmonic distortion rate, temperature gradient matrix, and historical fault time series are calculated to output the fault probability value. The dynamic balancing controller adjusts the working status of each unit in the hybrid energy storage module according to the fault probability value. Compared with the existing technology, this intelligent fault prediction and dynamic balancing control function can predict faults in advance and take measures in advance, such as adjusting the output power of the energy storage unit or performing charge and discharge control, to avoid the expansion and spread of faults, effectively improving the reliability and stability of the transmission line energy storage device and reducing the impact of faults on the transmission system.

[0048] (3) High-efficiency heat dissipation design: The composite heat dissipation shell is wrapped around the outside of the hybrid energy storage module, and the internal bionic air duct is tightly fitted with the surface of the energy storage unit. The heat dissipation efficiency is dynamically adjusted through temperature sensor signal feedback. The phase change material layer is a paraffin-expanded graphite-boron nitride composite material with a thermal conductivity of ≥18W / m·K, which can quickly conduct and dissipate the heat generated by the energy storage unit. The bionic air duct structure has a shark skin non-smooth surface, which can optimize airflow distribution and enhance the heat dissipation effect. Compared with existing technologies, this heat dissipation design can not only effectively solve the problem of heat accumulation during long-term operation of the energy storage device, but also dynamically adjust the heat dissipation efficiency according to actual working conditions, avoiding energy waste, ensuring that the energy storage device can operate stably under different environmental conditions, extending the service life of the energy storage device, and reducing maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a structural diagram of a new type of power supply energy storage device for power transmission lines of the present invention;

[0050] Figure 2 This is the training flowchart for the improved whale optimization algorithm. DETAILED DESCRIPTION

[0051] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0052] Unless otherwise defined, technical or scientific terms used in the present invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0053] Example 1

[0054] like Figure 1-2 The present invention provides a new type of power supply energy storage device for a transmission line, including: a hybrid energy storage module, an embedded fault prediction unit, a dynamic balancing controller and a composite heat dissipation shell.

[0055] The hybrid energy storage module consists of all-vanadium liquid flow battery units, solid-state lithium capacitor units and supercapacitor units connected in parallel through a bidirectional DC-DC converter, and the output ends of each unit are electrically connected through a DC bus.

[0056] The all-vanadium liquid flow battery unit consists of multiple single cell stacks. Its electrolyte uses a high-purity vanadium ion solution. The positive electrode electrolyte and the negative electrode electrolyte are stored in two independent storage tanks, each with a volume of 0.5 cubic meters. The electrolyte circulates between the battery stack and the storage tank through a pump circulation system to ensure normal charging and discharging of the battery.

[0057] In actual operation, when the transmission line is in a normal power supply state, the all-vanadium liquid flow battery unit is connected to the DC bus through a bidirectional DC-DC converter, supplying power to the grid with stable power. Its output power can be adjusted in the range of 0-50kW according to the needs of the grid.

[0058] The structure of the solid-state lithium capacitor unit includes: the positive electrode material is single crystal LiNi 0.8 Mn 0.1 Co 0.1 O2, specific capacity ≥ 200mAh / g; negative electrode material is silicon carbon nanowire array, diameter ≤ 80nm, loading ≥ 4mg / cm 2 ; The solid electrolyte layer is Li 1.5 Al 0.5 Ge 1.5 (PO4)3 film, thickness ≤ 20μm, ionic conductivity ≥ 1mS / cm; the battery cells are connected by copper-aluminum composite foil, and the gaps between adjacent battery cells are filled with aerogel insulation layer with a thermal conductivity coefficient ≤ 0.02W / m·K.

[0059] The supercapacitor uses a carbon-based double-layer capacitor, which uses activated carbon with a high specific surface area as the electrode material. It is formed by impregnating the electrolyte and then winding it, and has good rate performance and cycle life.

[0060] Multiple supercapacitors are directly connected to the DC bus after suppressing high-frequency oscillations through distributed micro-inductors. With an inductance of 20nH ±5%, they effectively filter out high-frequency noise and improve power quality. In extreme situations such as transmission line short-circuits, the supercapacitor units can deliver up to 200kW of power within 1 millisecond, providing strong support for power compensation at the moment of failure.

[0061] The bidirectional DC-DC converter utilizes a triple-interleaved parallel topology, with each parallel branch switching at 100kHz. Voltage conversion and electrical isolation are achieved through a high-frequency transformer. With an efficiency of ≥97% and a ripple of ≤1%, it effectively ensures energy transfer efficiency and voltage stability between different energy storage units.

[0062] The converter's control circuit is implemented using a digital signal processor (DSP), which has a fast response speed and precise control accuracy. It can complete the sampling and control of input and output voltages and currents within 10 microseconds, realizing real-time dynamic management of the hybrid energy storage module.

[0063] The circuit topology of the hybrid energy storage module includes: a triple interleaved parallel DC-DC converter between the all-vanadium liquid flow battery unit and the solid-state lithium capacitor unit, with an efficiency of ≥97% and a ripple of ≤1%; the supercapacitor unit is directly connected to the DC bus and uses distributed micro-inductors to suppress high-frequency oscillations, with an inductance of 20nH±5%; redundant switching channels are set between each unit, with a switching response time of ≤100μs, and the switching trigger signal is transmitted by a dynamic balancing controller through an optocoupler isolation circuit.

[0064] The embedded fault prediction unit is connected to the hybrid energy storage module through a data bus, and includes a current harmonic analysis module, a temperature field reconstruction module and a neural network accelerator. The output end of the current harmonic analysis module is connected to the input end signal of the neural network accelerator.

[0065] The current harmonic analysis module collects line current signals in real time and extracts characteristic harmonic spectrum data through a programmable FIR filter group. Its center frequency is dynamically configured to 6kHz, 12kHz, and 18kHz, and the extracted characteristic harmonic spectrum data is transmitted to the neural network accelerator through the SPI interface.

[0066] The update rules for the FIR filter group coefficients of the current harmonic analysis module are as follows:

[0067] Based on historical fault data, the center frequency is dynamically adjusted, and the update formula is:

[0068] ;

[0069] in, represents the new center frequency, represents the old center frequency, represents the learning rate, Indicates the The amplitude of the subharmonics, Indicates the The frequency corresponding to the subharmonic;

[0070] The filter coefficients are updated online via the JTAG interface every 24 hours;

[0071] When updating, a sliding window mechanism is used to select the latest 100,000 groups of sampling points as the update data window.

[0072] Temperature field reconstruction module:

[0073] Distributed fiber optic sensors, installed along key locations of transmission lines and energy storage devices, utilize fiber Bragg grating (FBG) sensors, which have a temperature measurement range of -40°C to 150°C, an accuracy of ±1°C, and a spatial resolution of ≤1m. A wavelength demodulator demodulates the optical signal reflected by the FBG sensors to obtain temperature data at each measurement point.

[0074] The temperature field reconstruction algorithm uses finite element analysis to construct a temperature distribution model for the entire energy storage device and transmission line based on measured discrete temperature data. During energy storage device operation, if the temperature of a storage cell rises abnormally, the temperature field reconstruction module accurately identifies the hotspot and sends the temperature data to the neural network accelerator for comprehensive fault risk assessment.

[0075] Neural Network Accelerator:

[0076] The specific structural parameters of the LSTM-TCN hybrid model are as follows: the LSTM layer contains 64 neurons, the TCN layer contains 3 convolutional blocks, each convolutional block has a kernel size of 3 and a number of channels of 32. The model has an input dimension of 100 and an output dimension of 1, which is the fault probability value.

[0077] The weights of the LSTM-TCN hybrid model are trained using the improved whale optimization algorithm:

[0078] Step S1: Initialize the position of the whale group. Each whale represents a set of connection weights and bias parameters of the LSTM-TCN hybrid model.

[0079] Step S2: Define the fitness function, and the calculation formula is:

[0080] ;

[0081] in, represents the fitness value, represents the harmonic mean of the model’s prediction accuracy and recall, represents the time required for the model to calculate the output failure probability value, Indicates the energy consumption during the model operation;

[0082] Step S3: Iteratively update the position of the whale group:

[0083] Calculate the fitness value of each whale;

[0084] Select the best whale based on fitness value;

[0085] Simulate whale encirclement, spiral updating, and prey search behaviors to update the whale group's position;

[0086] The update formula for surrounding the prey is:

[0087] ;

[0088] ;

[0089] in, Represents the distance vector between the current whale and the prey, which is used to determine the position update direction and distance of the whale when surrounding the prey. 、 represents the coefficient vector, , , Represents a control parameter that decreases linearly with the number of iterations and is used to balance global and local search capabilities. 、 Represents an independent standard random vector, used to increase the randomness and diversity of the search, represents the location of the prey (optimal solution), Indicates the current whale's position. Indicates that whales are The position is updated at all times based on the strategy of surrounding the prey;

[0090] The spiral update position formula is:

[0091] ;

[0092] ;

[0093] in, Indicates that whales are The updated position of the spiral path strategy at each moment, represents the spiral shape parameter, which determines the tightness of the spiral. Represents a random vector with values ​​in the range [-1, 1], used to control the direction of the spiral motion. represents the distance vector between the whale's current position and the prey's position, represents the base of natural logarithms, Represents element-wise multiplication.

[0094] Step S4: Introduce gradient information in the iterative process, combine the gradient descent method, calculate the gradient of the loss function with respect to the model parameters, and adjust the parameters along the gradient direction when the whale position is updated;

[0095] Step S5: In each iteration, normalize the model weights;

[0096] Step S6: Repeat steps S2 to S5 until the maximum number of iterations is reached or the convergence condition is met, and output the connection weights and bias parameters of the final optimized LSTM-TCN hybrid model.

[0097] The dynamic balancing controller is connected to the bidirectional DC-DC converter in the hybrid energy storage module, the active balancing circuit of the solid-state lithium capacitor unit, and the microfluidic pump of the all-vanadium liquid flow battery unit through multiple PWM signal lines, receives the fault probability value output by the neural network accelerator and generates control instructions.

[0098] The dynamic balancing controller receives the fault probability value output by the neural network accelerator. When the fault probability value is ≥0.6, it sends a three-level pre-charge instruction to the hybrid energy storage module to control the operating status of the supercapacitor unit, solid-state lithium capacitor unit, and all-vanadium liquid flow battery unit. The PWM duty cycle of the pre-charge instruction and the fault probability value satisfy a linear relationship:

[0099] ;

[0100] in, Indicates the PWM duty cycle of the precharge instruction, Represents the failure probability value.

[0101] The composite heat dissipation shell is wrapped around the outside of the hybrid energy storage module, and the internal bionic air duct is tightly fitted to the surface of the hybrid energy storage module. The heat dissipation efficiency is dynamically adjusted through temperature sensor signal feedback.

[0102] The design of the composite heat dissipation housing includes: a phase change material layer made of paraffin wax-expanded graphite-boron nitride composite material with a phase change latent heat of ≥200 J / g, a thermal conductivity of ≥18 W / m·K, and a thickness of 2.5 mm ± 0.1 mm; a bionic air duct structure with a shark skin non-smooth surface, a groove depth of 50 μm ± 5 μm, a spacing of 200 μm, and an inclination angle of 55° ± 2°; and heat dissipation efficiency adjustment logic that meets the following requirements:

[0103] ;

[0104] in, Indicates the heat dissipation efficiency, Indicates the real-time temperature in °C.

[0105] The adjustment instructions are sent to the cooling fan and semiconductor cooling plate through the CAN bus to achieve dynamic adjustment of the cooling efficiency.

[0106] Therefore, the present invention adopts the above-mentioned new power supply energy storage device for transmission lines, which significantly improves the performance, reliability and safety of the energy storage device, extends its service life, reduces maintenance costs, and is suitable for transmission lines with complex working conditions.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A power transmission line power storage device, characterized in that: include: Hybrid energy storage module, embedded fault prediction unit, dynamic balancing controller and composite heat dissipation housing; The hybrid energy storage module consists of all-vanadium liquid flow battery units, solid-state lithium capacitor units and supercapacitor units connected in parallel through a bidirectional DC-DC converter. The output ends of each unit are electrically connected through a DC bus. The embedded fault prediction unit is connected to the hybrid energy storage module through a data bus, and includes a current harmonic analysis module, a temperature field reconstruction module, and a neural network accelerator. The output of the current harmonic analysis module is connected to the input of the neural network accelerator. The dynamic balancing controller is connected to the bidirectional DC-DC converter in the hybrid energy storage module, the active balancing circuit of the solid-state lithium capacitor unit, and the microfluidic pump of the all-vanadium liquid flow battery unit through multiple PWM signal lines. It receives the fault probability value output by the neural network accelerator and generates control instructions. The composite heat dissipation shell is wrapped around the outside of the hybrid energy storage module, and the internal bionic air duct is bonded to the surface of the hybrid energy storage module. The heat dissipation efficiency is dynamically adjusted through temperature sensor signal feedback; The neural network accelerator has a built-in LSTM-TCN hybrid model that calculates the input harmonic distortion rate, temperature gradient matrix, and historical fault time series, and outputs a fault probability value. The weights of the LSTM-TCN hybrid model are trained using the improved whale optimization algorithm: Step S1: Initialize the position of the whale group. Each whale represents a set of connection weights and bias parameters of the LSTM-TCN hybrid model. Step S2: Define the fitness function, and the calculation formula is: ； in, represents the fitness value, represents the harmonic mean of the model’s prediction accuracy and recall, represents the time required for the model to calculate the output failure probability value, Indicates the energy consumption during the model operation; Step S3: Iteratively update the position of the whale group: Calculate the fitness value of each whale; Select the best whale based on fitness value; Simulate whale encirclement, spiral updating, and prey search behaviors to update the whale group's position; Step S4: Introduce gradient information in the iterative process, combine the gradient descent method, calculate the gradient of the loss function with respect to the model parameters, and adjust the parameters along the gradient direction when the whale position is updated; Step S5: In each iteration, normalize the model weights; Step S6: Repeat steps S2 to S5 until the maximum number of iterations is reached or the convergence condition is met, and output the connection weights and bias parameters of the final optimized LSTM-TCN hybrid model; The dynamic balancing controller receives the fault probability value output by the neural network accelerator. When the fault probability value is ≥0.6, it sends a three-level pre-charge instruction to the hybrid energy storage module to control the operating status of the supercapacitor unit, solid-state lithium capacitor unit, and all-vanadium liquid flow battery unit. The PWM duty cycle of the pre-charge instruction and the fault probability value satisfy a linear relationship: ； in, Indicates the PWM duty cycle of the precharge instruction, represents the failure probability value; The design of the composite heat dissipation housing includes: a phase change material layer made of paraffin wax-expanded graphite-boron nitride composite material with a phase change latent heat of ≥200 J / g, a thermal conductivity of ≥18 W / m·K, and a thickness of 2.5 mm ± 0.1 mm; a bionic air duct structure with a shark skin non-smooth surface, a groove depth of 50 μm ± 5 μm, a spacing of 200 μm, and an inclination angle of 55° ± 2°; and heat dissipation efficiency adjustment logic that meets the following requirements: ； in, Indicates the heat dissipation efficiency, Indicates the real-time temperature in °C. Represents the base of natural logarithms.

2. A power transmission line power supply energy storage device according to claim 1, characterized in that: The structure of the solid-state lithium capacitor unit includes: the positive electrode material is single crystal LiNi 0.8 Mn 0.1 Co 0.1 O2, specific capacity ≥ 200mAh / g; negative electrode material is silicon carbon nanowire array, diameter ≤ 80nm, loading ≥ 4mg / cm 2 ; The solid electrolyte layer is Li 1.5 Al 0.5 Ge 1.5 (PO4)3 film, thickness ≤ 20μm, ionic conductivity ≥ 1mS / cm; the battery cells are connected by copper-aluminum composite foil, and the gaps between adjacent battery cells are filled with aerogel insulation layer with a thermal conductivity coefficient ≤ 0.02W / m·K.

3. A power transmission line power supply energy storage device according to claim 1, characterized in that: The circuit topology of the hybrid energy storage module includes: a triple interleaved parallel DC-DC converter between the all-vanadium liquid flow battery unit and the solid-state lithium capacitor unit, with an efficiency of ≥97% and a ripple of ≤1%; the supercapacitor unit is directly connected to the DC bus and uses distributed micro-inductors to suppress high-frequency oscillations, with an inductance of 20nH±5%; redundant switching channels are set between each unit, with a switching response time of ≤100μs, and the switching trigger signal is transmitted by a dynamic balancing controller through an optocoupler isolation circuit.

4. A power transmission line power supply energy storage device according to claim 1, characterized in that: The update rules for the FIR filter group coefficients of the current harmonic analysis module are as follows: Based on historical fault data, the center frequency is dynamically adjusted, and the update formula is: ； in, represents the new center frequency, represents the old center frequency, represents the learning rate, Indicates the The amplitude of the subharmonics, Indicates the The frequency corresponding to the subharmonic; The filter coefficients are updated online via the JTAG interface every 24 hours; When updating, a sliding window mechanism is used to select the latest 100,000 groups of sampling points as the update data window.

5. The power transmission line power supply energy storage device according to claim 1, characterized in that: The current harmonic analysis module collects line current signals in real time and extracts characteristic harmonic spectrum data through a programmable FIR filter group. Its center frequency is dynamically configured to 6kHz, 12kHz, and 18kHz, and the extracted characteristic harmonic spectrum data is transmitted to the neural network accelerator through the SPI interface.

6. The power transmission line power supply energy storage device according to claim 1, characterized in that: The temperature field reconstruction module collects temperature data through distributed optical fiber sensors and sends the collected temperature data to the neural network accelerator via the I2C bus.

Images