Storage battery on-line monitoring and remote capacity checking system

Abstract

The invention belongs to the technical field of storage battery monitoring systems, and particularly relates to a storage battery on-line monitoring and remote capacity checking system, which comprises a sensing architecture layer for independently configuring a distributed integrated terminal module in each storage battery unit through a multi-physics field monitoring unit and monitoring the temperature, internal resistance and deformation of a storage battery pole; the analysis architecture layer is used for constructing a health evaluation module based on a fusion ampere-hour integral method, an internal resistance correction model and deep learning prediction to predict the health of the storage battery, and establishing an abnormal judgment strategy based on temperature gradient change and rapid capacity change quantity to carry out early warning on the safety of the storage battery; and the execution architecture layer integrates the three-level IGBT inversion module and the dynamic compensation module to carry out capacity checking on the storage battery, and configures an on-line on-load capacity checking mode and an off-line deep discharge mode. According to the invention, dangers and equipment power-off risks caused by a traditional capacity checking means can be eliminated, an automatic operation process is realized, the detection precision is high, and'monitoring-diagnosis-capacity checking 'closed-loop management is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery monitoring systems, and particularly relates to an on-line battery monitoring and remote capacity verification system. Background Art

[0002] In power grid substations, the judgment of the state of DC battery packs mainly relies on manual periodic discharge tests. Through annual verification discharges, it is determined whether the capacity of the battery pack is above 80% of the nominal capacity, and batteries with insufficient capacity are verified; through quarterly on-load discharges, open-circuit phenomena of the batteries are verified. The battery packs in substations are generally equipped with on-line battery monitoring mainly based on voltage monitoring, but it has no effect on judging the state of the batteries during operation.

[0003] At present, the verification discharge takes a long time, basically using resistive loads. The discharge of the battery pack is converted into heat and dissipated. There must be someone on-site during discharge, and the ventilation in the battery room must be turned on. Under the current situation of the continuous expansion of the power grid scale, the power grid maintenance department does not have enough personnel to carry out this work, and the nuclear discharge process is relatively dangerous, with a risk of interruption of DC system power supply.

[0004] Secondly, the existing monitoring system is not perfect. The internal resistance detection is nearly fixed and insufficient, the temperature field monitoring is missing, and at the same time, the physical state monitoring is blank. There is a lack of effective detection means for mechanical failures such as battery deformation and leakage. At the same time, the system intelligence level is low, there are data islands, the monitoring system lacks data linkage, and the monitoring error is relatively large. Summary of the Invention

[0005] The purpose of the present invention is to provide an on-line battery monitoring and remote capacity verification system, which can eliminate the dangers and equipment power-off risks brought by traditional capacity verification means, and has an automated operation process, a high level of intelligence, a high detection accuracy, and realizes a closed-loop management of "monitoring - diagnosis - capacity verification".

[0006] The technical solutions adopted by the present invention are specifically as follows: An on-line battery monitoring and remote capacity verification system, comprising: A perception architecture layer, which independently configures a distributed integration terminal module in each battery unit through a multi-physical field monitoring unit for monitoring the temperature, internal resistance and deformation of the battery terminal posts; An analysis architecture layer, which constructs a health assessment module to predict the health of the battery based on the integrated ampere-hour integration method, internal resistance correction model and deep learning prediction, and establishes an abnormal judgment strategy based on the temperature gradient change and the sharp change in capacity to give an early warning of the battery safety, forming a safety early warning module; An execution architecture layer, which integrates a three-level IGBT inverter module and a dynamic compensation module to perform capacity verification on the battery, and configures an on-line on-load capacity verification mode and an off-line deep discharge mode.

[0007] As a preferred solution, the multi-physical field monitoring unit specifically includes a voltage sensor unit, a temperature monitoring unit, an internal resistance detection unit, and a deformation monitoring unit; Among them; The voltage sensing unit uses a voltage sensor with an accuracy of ±0.05% and is configured on each battery unit by a four-wire high-precision voltage acquisition scheme; The temperature monitoring unit uses a thin-film sensor and is directly attached to the surface of the battery terminal post, and a five-point temperature measurement array is arranged at the center and four corners of the battery case; The internal resistance detection unit uses the AC injection method and is configured with a constant current source, achieving a resolution of 1 μΩ through a signal conditioning circuit, and the measurement period is 60 seconds / time; The deformation monitoring unit includes resistance strain gauges arranged in an array on the side of the battery case, and uses a Wheatstone bridge structure to configure temperature compensation gauges and eliminate the influence of ambient temperature.

[0008] As a preferred solution, the calculation formula for the amount of deformation in the deformation monitoring unit is as follows: ΔL / L = (ΔR / R) / (X×K); Where, ΔL / L represents the axial strain, ΔR / R represents the relative change in resistance, K is the sensitivity coefficient, and X is the number of resistance strain gauges; The axial strain is calculated using the component difference to exclude the influence of uniform expansion, and three-level alarm thresholds are set: Warning: Strain value > 200 με; Alarm: Strain difference between adjacent units > 50 με; Emergency: Strain gradient change rate > 10 με / min.

[0009] As a preferred solution, the ampere-hour integration method calculates the battery capacity by real-time collecting the charge and discharge current and using a piecewise integration algorithm. The specific formula is as follows: ; Where, is the true capacity of the battery, is the real-time current, negative for discharge and positive for charge, is the integration time window, is the temperature compensation factor, is the state of charge correction coefficient.

[0010] As a preferred solution, the internal resistance correction model uses a dynamic calibration algorithm. The specific formula is as follows: ; Where, is the current internal resistance value, is the initial internal resistance reference value, is the temperature coefficient, is the temperature change, is the capacity attenuation coefficient, is the current actual capacity of the battery, is the rated capacity of the battery.

[0011] As a preferred solution, the deep learning prediction outputs the health state prediction value and the remaining service life prediction value by building an LSTM neural network model and inputting the time-series voltage fluctuation, the temperature gradient distribution, and the historical capacity attenuation curve.

[0012] As a preferred solution, the three-level IGBT inverter module includes a three-level IGBT inverter unit and a grid connection control unit; wherein; The level IGBT inverter unit adopts a neutral point clamped three-level topology, configures an IGBT module, and the output filter adopts an LCL structure; The grid connection control unit uses an improved phase-locked loop technology to achieve grid voltage synchronization and maintains the stability of the bus voltage by deploying a reactive power compensation algorithm.

[0013] As a preferred solution, the dynamic compensation module dynamically adjusts the modulation ratio by establishing a double-loop structure of a current inner loop and a voltage outer loop and based on dq-axis decoupling control to compensate for the line impedance fluctuation.

[0014] As a preferred solution, the online on-load capacity verification mode performs inverter control by connecting a bidirectional DC / AC module in parallel with the system bus, and collects the bus voltage in real time to automatically adjust the discharge current; The off-line deep discharge mode switches the battery pack to be tested to an independent discharge circuit by configuring a bipolar STS switch and adopts a interleaved parallel Buck circuit to control the ripple current.

[0015] As a preferred solution, the online on-load capacity verification mode and the off-line deep discharge mode are switched by adopting an automatic identification algorithm based on the health state value of the battery pack; wherein, the health state value of the battery pack is the true capacity of the battery and the current internal resistance value, and the automatic identification algorithm is the ampere-hour integration method and the dynamic calibration algorithm.

[0016] The technical effects achieved by the present invention are: The present invention replaces the traditional manual capacity verification with an online monitoring and remote capacity verification system, eliminates the safety hazards of traditional capacity verification operations, and realizes on-load capacity verification through a parallel inverter device through an online / off-line hybrid capacity verification mode, ensuring continuous power supply of the DC system and avoiding the equipment power-off risk caused by traditional capacity verification operations.

[0017] Through an automated operation process, the present invention realizes the closed-loop management of "monitoring - diagnosis - capacity verification", greatly reduces the capacity verification time of a single battery pack, and also greatly reduces manual intervention, saving labor costs. By combining the LSTM neural network with the SOH evaluation model, the capacity estimation error is greatly reduced, the internal resistance detection accuracy is improved, and the battery deterioration can be predicted in advance for timely maintenance. At the same time, mechanical failures such as battery deformation and leakage can be effectively detected, with a high level of intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a schematic structural diagram of the system according to an embodiment of the present invention; Figure 2 is a schematic structural diagram of the hardware system in an embodiment of the present invention; Figure 3 is a schematic flow structural diagram of the multi-source data fusion evaluation model in an embodiment of the present invention; Figure 4 is a schematic flow structural diagram of the abnormal judgment strategy system in an embodiment of the present invention; Figure 5 is a schematic flow structural diagram of the intelligent capacity verification host construction in an embodiment of the present invention; Figure 6 is a schematic flow structural diagram of the dual-mode switching in an embodiment of the present invention; Figure 7 is a schematic work flow structural diagram of the capacity verification in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the purpose and advantages of the present invention clearer, the present invention will be specifically described below in conjunction with embodiments. It should be understood that the following text is only used to describe one or several specific embodiments of the present invention, and does not strictly limit the specific scope of protection claimed by the present invention.

[0020] As Figures 1 - 7 shown, the on-line monitoring and remote capacity verification system for storage batteries realizes the closed-loop management of "monitoring - diagnosis - capacity verification" of storage batteries by constructing a three-layer architecture of "perception - analysis - execution", and at the same time combines the LSTM neural network with the SOH evaluation model to ensure the detection accuracy of the internal resistance of storage batteries and be able to predict the deterioration of batteries in advance, specifically including: I. Construction of the perception layer 1. Construction of the hardware system (1) Deployment of voltage sensors Adopt a four-wire high-precision voltage acquisition scheme, and independently configure a voltage sensor with an accuracy of ±0.05% for each storage battery; the connection of the electrode posts uses gold-plated spring probes, with a contact resistance <0.1 mΩ, and configure an EMI filter circuit to eliminate ripple interference.

[0021] Among them, the four-wire high-precision voltage acquisition scheme is specifically to separate the current transmission and voltage measurement paths, eliminate the influence of wire resistance on the measurement accuracy, and is suitable for the accurate acquisition of microvolt-level voltage signals. Its principle is to apply a constant current to the battery under test through two wires of the current excitation loop, and use the voltage detection loop to directly measure the true voltage between the two poles of the battery with the other two wires. This scheme can eliminate interference such as contact resistance and cable resistance, and the accuracy can reach ±0.05%.

[0022] It should be noted that a spring-type gold-plated probe is used to directly contact the battery terminal post (contact resistance <0.1 mΩ), and at the same time, a low-noise operational amplifier (such as AD8421) is configured to suppress common-mode interference, and an RC filter network is added at the input end using an EMI filter circuit to eliminate high-frequency ripple interference.

[0023] (2)Construct a temperature monitoring network To monitor the temperature of the terminal post: install a thin-film PT1000 sensor (class A accuracy, ±0.1 °C) and directly attach it to the surface of the terminal post; To monitor the temperature of the battery case: arrange a 5-point temperature measurement array at the center / corners of the battery case and adopt an I2C bus cascaded topology.

[0024] (3)Internal resistance detection module Adopt the 1 kHz AC injection method and configure a ±5 mA constant current source (accuracy 0.1%); specifically as follows: Use the REF200 chip to construct a ±5 mA constant current source, and achieve 0.1% accuracy through a temperature compensation circuit; and configure a four-wire output: the current output terminals (Force+ / Force-) and the voltage detection terminals (Sense+ / Sense-) are independently wired to eliminate the influence of wire impedance; use the AD9833 signal generator to generate a 1 kHz sine wave, and the phase noise < -100 dBc / Hz; suppress the injected current ripple, and connect a 0.1 μF ceramic capacitor + 10 Ω damping resistor combination in parallel at the output end.

[0025] Construct a signal conditioning circuit: including a pre-amplification module (gain 100 times) + band-pass filter (Q value 50) + synchronous demodulation module; among them, the pre-amplification module selects the AD620 instrumentation amplifier and configures a gain of 100 times; the common-mode rejection ratio (CMRR) > 100 dB, and the input impedance > 10 GΩ.

[0026] Secondly, set a second-order active filter (center frequency 1 kHz, Q = 50); the cut-off frequency of the high-pass filter is 980 Hz; the cut-off frequency of the low-pass filter is 1020 Hz; use the AD8675 operational amplifier, and the temperature drift < 0.5 μV / °C; at the same time, use the AD630 balanced modulator to achieve phase-sensitive detection.

[0027] In this implementation, a resolution of 1 μΩ can be achieved, the measurement period is 60 seconds per time, and the influence of contact impedance is eliminated.

[0028] (4)Deformation monitoring module Strain gauge array: A 4×4 resistor strain gauge network (grid length 2 mm, sensitivity coefficient 2.0) is arranged on the side of the battery case; and a Wheatstone bridge structure is adopted, and temperature compensation gauges are configured to eliminate the influence of ambient temperature; Calculation of deformation amount: ΔL / L = (ΔR / R) / (X×K), and the detection accuracy reaches ±0.01 mm; where, ΔL / L represents the axial strain, ΔR / R represents the relative change in resistance, K is the sensitivity coefficient, and X is the number of resistor strain gauges.

[0029] In this embodiment, a 4×4 resistor strain gauge network is adopted, X = 4, and the specific calculation formula is ΔL / L = (ΔR / R) / (4×K).

[0030] It should be noted that the axial strain is calculated using the component difference to exclude the influence of uniform expansion, and three-level alarm thresholds are set: Early warning: Strain value > 200 με; Alarm: Strain difference between adjacent units > 50 με; Emergency: Strain gradient change rate > 10 με / min; When the specified threshold is reached, the system will send corresponding early warning information to prompt the staff.

[0031] 2. Distributed integration terminal By configuring an independent acquisition module for each battery unit, and configuring an STM32G473 (built-in 24-bit Σ-Δ ADC) + AD8421 instrumentation amplifier core chip, and supporting dual modes of battery power supply (3.6 - 15 VDC wide voltage input) and external power supply; at the same time, the acquired voltage signal is directly sampled after RC low-pass filtering (cutoff frequency 100 Hz); the temperature signal is driven by a constant current source through a PT1000, and sampled by a 24-bit ADC; the internal resistance signal calculates the real part impedance after phase-sensitive detection, and eliminates the influence of capacitive components.

[0032] It should be noted that a timestamp alignment mechanism can be established in this system to ensure the time scale synchronization of voltage / temperature / internal resistance data.

[0033] II. Construction of the analysis layer 1. Multi-source data fusion evaluation model (1)Ampere-hour integration method (Ah) By collecting the charge and discharge current in real time, and using a piecewise integration algorithm to calculate the battery capacity, the specific formula is as follows: ; Among them, is the true capacity of the battery, is the real-time current, negative for discharging and positive for charging, is the integration time window, is the temperature compensation factor, is the state of charge correction coefficient.

[0034] During the charging and discharging processes of the battery, the current and time are monitored in real time, and then the charge quantity is calculated using the above formula. When performing the capacity verification test on the battery, during the entire process from the start of discharging to the end of discharging, the discharging current and time are continuously recorded, and the actually discharged charge quantity is obtained through integral calculation to evaluate the current actual capacity of the battery.

[0035] (2)Internal resistance correction model Adopt a dynamic calibration algorithm, and the specific formula is as follows: ; Among them, is the current internal resistance value, is the initial internal resistance reference value, is the temperature coefficient, is the temperature change amount, is the capacity attenuation coefficient, is the current actual capacity of the battery, is the rated capacity of the battery.

[0036] In this embodiment, the specific application steps are as follows: Step 1: Regularly measure the internal resistance of the battery at the current temperature , and at the same time record the ambient temperature and the initial temperature (generally, the temperature at the time of battery production or in a brand-new state is used as ), and calculate the temperature change amount .

[0037] Step 2: Given the initial internal resistance and the temperature coefficient (for different types of batteries, their temperature coefficients are different, and generally, the data provided by the battery manufacturer or experimental determination can be used), use the internal resistance correction model to calculate the corrected internal resistance.

[0038] Step 3: Compare the corrected internal resistance with the standard internal resistance range when the battery is in good health state. If the corrected internal resistance exceeds the standard range, it indicates that the battery may have problems such as aging and internal faults, and further inspection and maintenance are required Secondly, the capacity calculated by the ampere-hour integration method can be combined with the internal resistance correction model to more accurately evaluate the state of health of the battery. As the battery is used, its internal resistance gradually increases and its capacity gradually decays. By comparing the currently calculated capacity with the nominal capacity of the battery and combining it with the change in internal resistance, the degree of battery aging can be judged.

[0039] Specifically, if the calculated actual capacity is significantly lower than the nominal capacity and the internal resistance also increases significantly, it indicates that the state of health of the battery is poor.

[0040] 2. Deep learning prediction Build an LSTM neural network model, and the input features include: Time-series voltage fluctuations (sampling rate 1Hz); Temperature gradient distribution (at three levels: terminal / post / ambient); Historical capacity decay curve.

[0041] Output: SOH (state of health) prediction value (error < 2%); Remaining useful life (RUL) estimate (confidence > 90%).

[0042] It should be noted that when calculating, this analysis layer can automatically adjust the algorithm weights according to the battery type (lead-acid / lithium-ion): among them, the internal resistance weight of lead-acid batteries accounts for 60%, and the capacity accounts for 30%; while the voltage balance weight of lithium-ion batteries is increased to 45%; to meet the real-time calculation requirements.

[0043] Secondly, during actual use, perform baseline parameter calibration, static internal resistance measurement (0.1C pulse discharge), and temperature sensor zero-point calibration every 24 hours.

[0044] In this embodiment, the SOH estimation of the battery based on LSTM is trained by extracting health features such as constant-current charging time, constant-voltage charging time, and polarization internal resistance as inputs and combining with the Oxford lithium-ion battery aging dataset. LSTM can effectively learn the battery capacity decay law and optimize the prediction accuracy by adjusting network parameters.

[0045] Secondly, in the prediction of the remaining life of the battery, LSTM constructs a time-series input sequence by analyzing parameters such as charge-discharge cycle times, voltage, and temperature. In this embodiment, the average absolute error of the prediction model based on LSTM can reach 0.05 years, which is better than traditional capacity decay models and state of health models.

[0046] Of course, further combining CNN to extract spatial features to form a CNN-LSTM hybrid model can further improve the prediction efficiency and accuracy.

[0047] In specific applications, the data calculated by the ampere-hour integration method is used as the input of the deep learning model. By leveraging the powerful learning ability of deep learning, the future capacity change, remaining service life, etc. of the battery are predicted. The deep learning model can learn the complex patterns and rules in the historical charge and discharge data of the battery, thereby improving the accuracy and reliability of the state of health assessment.

[0048] Furthermore, by analyzing a large amount of historical ampere-hour integration data and the corresponding battery state of health labels, a recurrent neural network (RNN) or long short-term memory network (LSTM) model can be trained to predict the future capacity decay trend of the battery.

[0049] 3. Abnormal Judgment Strategy System (1) Temperature Gradient Model Establish a three-dimensional thermal field analysis model: Threshold of the temperature rise rate of the terminal post: > 5°C / min (trigger a first-level alarm); Threshold of the temperature difference of the battery case: Temperature difference between adjacent single cells > 3°C (trigger a second-level alarm).

[0050] Thermal runaway prediction algorithm: ; Among them, : Temperature change amount (temperature difference between the terminal post and the battery case); : Time window (typical value 5 minutes); : Thermal conductivity of the material (0.8 - 1.2 for lead-acid batteries, 1.5 - 2.0 for lithium batteries).

[0051] (2) Capacity Sudden Drop Model Monitor the capacity decay rate in real time: Short-term mutation: Capacity drops > 5% within 5 minutes (trigger an emergency shutdown); Long-term trend: Monthly decay rate > 2% (trigger a maintenance reminder); It should be noted that in this embodiment, a sliding window algorithm (window size = 30 cycles) is used to filter noise.

[0052] III. Execution Layer Construction 1. The specific steps for building the intelligent capacity measurement host are as follows: Step 1. Construct a three-level IGBT inverter module Adopt a neutral-point clamped three-level topology, configure a 1200V IGBT module, and use an LCL structure for the output filter (inductance value 2mH + capacitance value 30μF) to achieve the goal of THD < 3%; at the same time, equip a water-cooled heat dissipation system (flow rate 5L / min) to control the IGBT junction temperature ≤ 85°C and ensure the continuous discharge ability of 0.5C.

[0053] Step 2: Build a dynamic compensation unit Establish a double-loop structure of current inner loop (PI controller) + voltage outer loop (droop control), with a sampling frequency of 10 kHz to achieve an accuracy of ±2%; at the same time, based on dq-axis decoupling control, dynamically adjust the modulation ratio (in the range of 0.8 - 1.2), and deploy a compensation algorithm to compensate for line impedance fluctuations.

[0054] Step 3: Construct a grid connection control strategy Adopt the phase-locked loop technology (PLL) to achieve grid voltage synchronization (phase difference < 0.5°); and deploy a reactive power compensation algorithm (Q-V droop control) to maintain the bus voltage stable within the range of ±2%.

[0055] It should be noted that the phase-locked loop technology (PLL) is specifically composed of a dual-mode phase detector and an adaptive filter; among them, the dual-mode phase detector is used to detect the phase difference and frequency difference simultaneously to shorten the locking time (such as adopting a Bang-Bang type + PFD hybrid architecture); secondly, the adaptive filter is used to dynamically adjust the loop bandwidth to balance noise suppression and response speed (such as introducing a moving average filter MAF).

[0056] Step 4: Establish a safety protection mechanism Set four-level protection trigger logic: overvoltage (±5%), overcurrent (±10%), overtemperature (65°C), harmonic exceedance (THD > 5%); and configure a redundant IGBT drive circuit with a fault switching time < 100 μs.

[0057] 2. The specific implementation steps of the dual-mode switching are as follows: Step 1: Online load-carrying and capacitance testing mode (discharge at 0.5C) (1) Parallel inverter control: Connect in parallel with the system bus through a bidirectional DC / AC module (efficiency > 95%) to achieve energy feedback to the grid; (2) Dynamic load matching: Real-time collect the bus voltage (accuracy ±0.5%) and automatically adjust the discharge current (adjustable from 0.1C to 0.5C).

[0058] Step 2: Offline deep discharge mode (discharge at 1.0C) (1) STS static switch: Configure a bipolar STS switch (switching time < 10 ms) to switch the battery pack to be tested to an independent discharge circuit; (2) High-current control: Adopt an interleaved parallel Buck circuit (4-way parallel) to control the ripple current within ±1%.

[0059] Step 3: Mode switching control logic (1) Automatic recognition algorithm: Based on the SOH value of the battery pack (enable the online mode when the capacity ≥ 80%, and force offline activation when < 80%); (2)Seamless transition strategy: Adopt pre-synchronization technology to adjust the output voltage phase of the inverter 3 seconds before the switch, achieving shock-free switching.

[0060] Step 4. Busbar design specifications Adopt a laminated busbar structure (spacing ≥ 5 mm), configure an RC absorption circuit (capacitor 0.1 μF + resistor 10 Ω); at the same time, the cross-sectional area of the copper bar in the discharge loop ≥ 50 mm², and the temperature rise control < 30 K (when the ambient temperature is 40 °C).

[0061] It should be noted that the online on-load capacity verification mode and the offline deep discharge mode are switched through an automatic recognition algorithm based on the state of health value of the battery pack; Among them, the state of health value of the battery pack is the true capacity and the current internal resistance of the battery, and the automatic recognition algorithm is the ampere-hour integration method and the dynamic calibration algorithm.

[0062] In the specific application process, as follows: I. Switching from the online on-load capacity verification mode to the offline deep discharge mode: (1)System instruction issuance: The operator issues a mode switching instruction through the visualization monitoring platform or the remote operation terminal at the application layer. After the control layer receives the instruction, it conducts a legality verification and a safety check on the instruction. After confirming that it is correct, it is ready to execute the switching operation.

[0063] (2)Load transfer and isolation: Start the intelligent switching cabinet (such as STS static switch) in the dual-mode switching device to gradually transfer some or all of the loads to the standby power supply to ensure that the load power supply is not affected during the switching process. For example, for important DC loads, the seamless transfer of the load can be completed within a few milliseconds through a fast-switching static switch.

[0064] When the load transfer is completed, isolate the battery pack from the online DC system and cut off its electrical connection with the busbar to ensure that it will not affect other devices during the offline deep discharge process.

[0065] (3)Parameter setting adjustment: According to the requirements of the offline deep discharge mode, adjust the parameter settings of the intelligent capacity verification host. Set the discharge current to 1.0C, and at the same time set key parameters such as the discharge duration and the cut-off voltage. The safety warning system synchronously updates the corresponding threshold settings, such as the temperature gradient threshold and the capacity sudden drop threshold, to meet the safety monitoring requirements in the offline deep discharge mode.

[0066] (4)Start the offline deep discharge: The control layer sends an instruction to start the offline deep discharge to the execution layer, and the intelligent capacity verification host starts to conduct a deep discharge operation on the battery pack according to the set parameters.

[0067] It should be noted that the sensing layer collects various parameters of the battery (such as voltage, current, temperature, etc.) in real time and transmits the data to the control layer for analysis and processing. The safety warning system continuously monitors the battery status. Once abnormal situations occur (such as too high temperature, rapid sudden drop in capacity, etc.), it immediately issues an alarm and takes corresponding protection measures.

[0068] II. Switching from the offline deep discharge mode to the online load-carrying capacity verification mode: (1) Discharge end judgment and preparation: When the offline deep discharge reaches the set termination condition (such as reaching the specified discharge duration or the battery voltage drops to the termination voltage), the intelligent capacity verification host automatically stops the discharge operation. The battery pack is given a short static time to allow the chemical reactions inside the battery to reach an equilibrium state. At the same time, the various parameters of the battery are checked again to evaluate whether the state of the battery is suitable for reconnecting to the online system.

[0069] (2) Charge recovery and parameter adjustment: According to the remaining capacity and state of the battery, start the charging program to recover the charge of the battery. During the charging process, a multi-stage charging strategy can be adopted, such as constant current charging, constant voltage charging, etc., to ensure that the battery is fully charged without overcharging. Adjust the parameter settings of the intelligent capacity verification host, set the discharge current to 0.5C, and at the same time update other relevant parameters (such as charging duration, charging voltage, etc.) to meet the requirements of the online load-carrying capacity verification mode.

[0070] (3) Reconnect to the online system: When the battery pack is charged and restored to a suitable state, the battery pack is reconnected to the DC system operating online through the intelligent switching cabinet in the dual-mode switching device. During the connection process, ensure that parameters such as voltage and polarity match to avoid generating impact current. Gradually transfer the load from the backup power supply back to the bus powered by the battery pack to achieve a smooth transition of the load.

[0071] (4) Start online load-carrying capacity verification: The control layer issues an instruction to start online load-carrying capacity verification, and the intelligent capacity verification host starts to perform online load-carrying capacity verification operations on the battery pack according to the set parameters. The sensing layer continuously collects various parameters of the battery and the system, the control layer analyzes the data in real time, and the safety warning system closely monitors the system status to ensure the safety and stability of the online load-carrying capacity verification process.

[0072] The above is only the preferred implementation mode of the present invention. It should be pointed out that for those of ordinary skill in the art in this technical field, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention. The structures, devices, and operation methods not specifically described and explained in the present invention are implemented according to the conventional means in this field without special description and limitation.

Claims

1. An on-line monitoring and remote capacity verification system for storage batteries, characterized in that Including: The perception architecture layer, where a distributed integrated terminal module is independently configured for each battery cell through a multi-physical field monitoring unit to monitor the temperature, internal resistance, and deformation of the battery terminal posts. The analysis architecture layer, where a health assessment module is constructed based on the integrated ampere-hour integration method, internal resistance correction model, and deep learning prediction to predict the health of the battery, and an abnormal judgment strategy is established based on the temperature gradient change and the sharp change in capacity to give early warnings for battery safety, forming a safety warning module. The execution architecture layer, where a three-level IGBT inverter module and a dynamic compensation module are integrated to perform capacity verification on the battery, and an online on-load capacity verification mode and an offline deep discharge mode are configured.

2. The on-line monitoring and remote capacity verification system for storage batteries according to claim 1, characterized in that: The multi-physical field monitoring unit specifically includes a voltage sensor unit, a temperature monitoring unit, an internal resistance detection unit, and a deformation monitoring unit. Among them; The voltage sensing unit uses a voltage sensor with an accuracy of ±0.05% and is configured on each battery cell using a four-wire high-precision voltage acquisition scheme. The temperature monitoring unit uses a thin-film sensor and is directly attached to the surface of the battery terminal post, and a five-point temperature measurement array is arranged at the center and four corners of the battery case. The internal resistance detection unit uses the AC injection method and is configured with a constant current source, achieving a resolution of 1 μΩ through a signal adjustment circuit, and the measurement period is 60 seconds per time. The deformation monitoring unit includes resistance strain gauges arranged in an array on the side of the battery case, and a temperature compensation gauge is configured using a Wheatstone bridge structure to eliminate the influence of ambient temperature.

3. The on-line monitoring and remote capacity verification system for storage batteries according to claim 2, wherein: The calculation formula for the amount of deformation in the deformation monitoring unit is as follows: ΔL / L = (ΔR / R) / (X×K); where, ΔL / L represents the axial strain, ΔR / R represents the relative change in resistance, K is the sensitivity coefficient, and X is the number of resistance strain gauges. The axial strain is calculated using the component difference to exclude the influence of uniform expansion, and three-level alarm thresholds are set: Early warning: Strain value > 200 με; Alarm: Strain difference between adjacent units > 50 με; Emergency: Strain gradient change rate > 10 με / min.

4. The on-line monitoring and remote capacity calibration system for storage batteries according to claim 1, characterized in that: The ampere-hour integration method calculates the battery capacity by real-time collecting the charge and discharge current and using a segmented integration algorithm. The specific formula is as follows: ； Among them, is the true capacity of the battery, is the real-time current, negative for discharging and positive for charging, is the integration time window, is the temperature compensation factor, is the state of charge correction coefficient.

5. The on-line monitoring and remote capacity calibration system for storage batteries according to claim 4, characterized in that: The internal resistance correction model uses a dynamic calibration algorithm. The specific formula is as follows: ； Among them, is the current internal resistance value, is the initial internal resistance reference value, is the temperature coefficient, is the temperature change amount, is the capacity attenuation coefficient, is the current actual capacity of the storage battery, is the rated capacity of the storage battery.

6. The on-line monitoring and remote capacity calibration system for storage batteries according to claim 1, characterized in that: The deep learning prediction outputs the predicted health status value and the predicted remaining service life value by building an LSTM neural network model and inputting the time-series voltage fluctuation, temperature gradient distribution, and historical capacity attenuation curve.

7. The on-line monitoring and remote capacity calibration system for storage batteries according to claim 1, wherein: The three-level IGBT inverter module includes a three-level IGBT inverter unit and a grid connection control unit. Among them; The level IGBT inverter unit uses a neutral point clamped three-level topology, is configured with an IGBT module, and the output filter uses an LCL structure. The grid connection control unit uses an improved phase-locked loop technology to achieve grid voltage synchronization and maintains the stability of the bus voltage by deploying a reactive power compensation algorithm.

8. The on-line monitoring and remote capacity calibration system for storage batteries according to claim 1, characterized in that: The dynamic compensation module dynamically adjusts the modulation ratio by establishing a double-loop structure of a current inner loop and a voltage outer loop and based on dq-axis decoupling control to compensate for the line impedance fluctuation.

9. The on-line monitoring and remote capacity verification system for storage batteries according to claim 4, characterized in that: The on-line load-carrying capacity verification mode performs inversion control by paralleling with the system bus through a bidirectional DC / AC module, and collects the bus voltage in real time to automatically adjust the discharge current; The off-line deep discharge mode configures a bipolar STS switch to switch the battery pack to be tested to an independent discharge circuit, and adopts an interleaved parallel Buck circuit to control the ripple current.

10. The on-line monitoring and remote capacity calibration system for storage batteries according to claim 9, wherein: The on-line load-carrying capacity verification mode and the off-line deep discharge mode perform mode switching by adopting an automatic recognition algorithm based on the health state value of the battery pack; Among them, the health state value of the battery pack is the true capacity and the current internal resistance of the battery, and the automatic recognition algorithm is the ampere-hour integration method and the dynamic calibration algorithm.

Images