A distributed liquid-cooled energy storage system intelligent management method and system

Abstract

The application discloses a kind of distributed liquid cooling energy storage system intelligent management method and system, comprising: identifying the sound wave reflection signal of sound wave reflection source type as bubble;Bubble static aggregation area is determined based on the distance between the position of bubble and bubble;The fluid retention degree index in bubble aggregation area is calculated, the fluid retention degree in bubble aggregation area is judged;Whether the local overheating risk exists in the bubble static aggregation area with the fluid retention degree being serious is judged;The temperature, bubble size and data of bubble static aggregation area in future preset time period are predicted, and the time point of local overheating risk is determined;The accuracy of bubble size prediction model, fluid retention degree index formula and local overheating risk assessment model is evaluated and optimized.The technical method provided by the present application significantly improves the accuracy of bubble monitoring, the timeliness of overheating risk assessment and the optimization efficiency of cooling effect in liquid cooling energy storage system, prolongs the service life of equipment and reduces maintenance cost.

Description

Technical Field

[0001] This invention relates to the field of next-generation information technology, and in particular to an intelligent management method and system for a distributed liquid-cooled energy storage system. Background Technology

[0002] With the growth of energy demand, especially in areas such as electric vehicles, high-efficiency data centers, and renewable energy storage, the demand for energy storage systems has increased dramatically. Liquid-cooled energy storage systems are widely used in these fields due to their excellent thermal management performance. Liquid cooling systems effectively control equipment temperature and maintain it at its optimal operating state through heat exchange between the coolant and the equipment, thereby improving the operating efficiency of the energy storage system. However, the effectiveness of liquid cooling systems is greatly affected by the problem of bubble accumulation. The presence of bubbles not only affects the flowability of the coolant but can also lead to localized overheating, and in severe cases, even equipment failure. Bubbles obstruct the normal flow of coolant, reduce cooling efficiency, and may even form static bubble accumulation areas, leading to localized overheating. Localized overheating not only affects the cooling effect but may also cause the energy storage equipment temperature to become too high, thus negatively impacting the system's reliability and lifespan. Therefore, timely detection, location, and handling of bubble accumulation areas are crucial for improving the performance of liquid-cooled energy storage systems. However, existing bubble detection methods often rely on traditional sensors, such as temperature and flow sensors. While these sensors can provide some monitoring information, their capabilities for bubble detection, location, and real-time tracking are very limited. These methods struggle to comprehensively and in real-time monitor the bubble state and cannot detect bubble accumulation problems in their early stages. Furthermore, traditional cooling management methods typically lack intelligent predictive mechanisms, failing to accurately predict bubble accumulation and temperature changes. This means that liquid cooling systems may not be able to respond promptly to overheating caused by bubble accumulation under high load conditions, leading to equipment failure or performance degradation. In summary, current liquid-cooled energy storage systems still face numerous technical challenges in bubble detection, location, overheating risk assessment, and dynamic prediction. Traditional monitoring methods cannot monitor bubble accumulation in real-time and comprehensively, and existing overheating risk assessment and management methods lack intelligent predictive capabilities, preventing energy storage systems from responding promptly to bubble accumulation or overheating risks. Therefore, how to accurately and in real-time monitor bubble changes, assess the overheating risk of static bubble accumulation areas, and implement dynamic maintenance measures has become a critical issue that urgently needs to be addressed in current liquid-cooled energy storage systems. Summary of the Invention

[0003] To address the problems existing in the prior art, this invention provides an intelligent management method and system for distributed liquid-cooled energy storage systems.

[0004] The first aspect of this invention provides an intelligent management method for a distributed liquid-cooled energy storage system, mainly comprising:

[0005] An ultrasonic beam is emitted by an ultrasonic sensor and the reflected sound wave signal is received by a directional array sensor. The spectral characteristics of the reflected sound wave signal are extracted using a short-time Fourier transform algorithm to identify the sound wave reflection source as a bubble.

[0006] The spectral characteristics of the acoustic wave reflection signal of the bubble as the acoustic wave reflection source are obtained, the bubble size is predicted, the bubble position is determined by time difference estimation, and the static bubble aggregation region is determined based on the bubble position and the distance between bubbles.

[0007] Based on the number and size of bubbles in the static bubble aggregation region, the bubble aggregation density in the static bubble aggregation region is determined, and the fluid retention index in the bubble aggregation region is calculated to determine the fluid retention degree in the bubble aggregation region.

[0008] Based on the load data of the energy storage system and the location and temperature data of the static bubble accumulation area with severe fluid stagnation, determine whether there is a risk of local overheating in the static bubble accumulation area with severe fluid stagnation, and formulate maintenance measures for local overheating risk.

[0009] Based on the temperature data, bubble size, and number of the static bubble accumulation area where there is no risk of local overheating, predict the temperature, bubble size, and data of the static bubble accumulation area within a preset time period in the future, and determine the time point when there is a risk of local overheating in the static bubble accumulation area.

[0010] Based on the local overheating risk data when implementing maintenance measures for the overheating risk of static bubble accumulation areas, the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model is evaluated and optimized.

[0011] Furthermore, the step of emitting an ultrasonic beam via an ultrasonic sensor and receiving the reflected sound wave signal via a directional array sensor, extracting the spectral characteristics of the reflected sound wave signal using a short-time Fourier transform algorithm, and identifying the reflected sound wave signal as a bubble as the source of the reflected sound wave includes:

[0012] An ultrasonic beam is emitted into the coolant of a distributed liquid-cooled energy storage system using an ultrasonic sensor, and the reflected sound wave signal is received by a directional array sensor. Based on the received reflected sound wave signal, an independent component analysis algorithm is used to separate the reflected sound wave signal from different sound wave sources. A short-time Fourier transform algorithm is used to convert the time-domain audio signal of the reflected sound wave signal into frequency-domain data, extract the spectral features of the reflected sound wave signal, and save them to a sound wave reflection monitoring database. The spectral features include frequency, peak frequency, and amplitude. Through the sound wave reflection monitoring database, the spectral features of the reflected sound wave signals from different sound wave sources are obtained, and the type of sound wave reflection source is labeled as a bubble. A support vector machine algorithm is used to train the model and build a bubble recognition model to identify the reflected sound wave signal of a bubble source.

[0013] Further, the step of acquiring the spectral characteristics of the acoustic wave reflection signal of a bubble as the acoustic wave reflection source, predicting the bubble size, determining the bubble position through time difference estimation, and determining the static bubble aggregation region based on the bubble position and the distance between bubbles includes:

[0014] Based on the real-time received acoustic wave reflection signals, a bubble recognition model is used to identify acoustic wave reflection signals whose source type is bubbles. The number of acoustic wave reflection signals whose source type is bubbles and the spectral characteristics of the corresponding acoustic wave reflection signals are obtained. A random forest regression algorithm is used to train the model and construct a bubble size prediction model to predict the bubble size. Based on the echo time of the bubble acoustic wave reflection signals obtained by the directional array sensor, the distance of the bubble relative to the sensor array is calculated through time difference estimation to determine the position of the bubble. The position changes of the bubbles are continuously monitored, and the distance between the bubbles is calculated based on the position changes of the bubbles. Based on the position changes of the bubbles and the distance between the bubbles, the static bubble aggregation area is determined.

[0015] Further, the step of determining the bubble aggregation density in the static bubble aggregation region based on the number and size of bubbles in the region, and calculating the fluid retention index within the region to determine the fluid retention degree, includes:

[0016] To obtain the number and size of bubbles within the static bubble aggregation region, use the bubble aggregation density factor formula. Calculate the bubble aggregation density ρ in the static bubble aggregation region. b ′, where N b It is the number of bubbles, D b It is the average size of the bubbles, I avg It is the average intensity of the sound wave reflection signal within the static aggregation region of the bubbles; based on the bubble size and the distance between the bubbles, the bubble interaction factor formula is used. Determine the bubble interaction factor α b′, where d b It is the minimum distance between bubbles, α b The influence of reaction bubble aggregation on flow; based on bubble aggregation density and bubble interaction factor, using the fluid retention index formula. The fluid retention index R within the bubble aggregation region is determined, where β and γ are adjustment factors obtained by fitting historical data, n is the index of the effect of bubble aggregation density on retention, used to control the rate of increase in retention as density increases, and ρ0′ is the baseline value of bubble aggregation density under normal operating conditions. Based on the fluid retention index within the bubble aggregation region and the preset index threshold, the fluid retention degree within the bubble aggregation region is judged, and the fluid retention degree includes severe, moderate, and ordinary.

[0017] Furthermore, based on the load data of the energy storage system and the location and temperature data of the bubble static accumulation area with severe fluid stagnation, it is determined whether there is a risk of local overheating in the bubble static accumulation area with severe fluid stagnation, and maintenance measures for local overheating risk are formulated, including:

[0018] The system acquires the location of static bubble accumulation areas, including pipe bends, joints, changes in location, and heat dissipation components. It obtains load data from the energy storage system via the battery management system and temperature data of severely stagnant static bubble accumulation areas using an infrared thermal imager. Load data includes current, voltage, and power. Based on the energy storage system's load data and the location and temperature data of severely stagnant static bubble accumulation areas, a recurrent neural network is used to train a model to construct a local overheating risk assessment model. Using the real-time acquired load data and the location and temperature data of severely stagnant static bubble accumulation areas, the local overheating risk assessment model is used to determine whether there is a risk of local overheating in these areas. If a local overheating risk is identified, maintenance measures are developed and implemented, including but not limited to adjusting the flow rate of the coolant within the energy storage system and adjusting the pressure of the liquid cooling system to remove bubbles from the static bubble accumulation areas.

[0019] Furthermore, the step of predicting the temperature, bubble size, and number of the static bubble accumulation area within a preset time period based on temperature data, bubble size, and number of the area without local overheating risk, and determining the time point when the static bubble accumulation area has a local overheating risk, includes:

[0020] If the fluid retention level in the static bubble accumulation area is moderate or average, the size and number of bubbles in the area are continuously acquired. A long short-term memory network is used to train the model, predicting the bubble size and number within a preset time period. This, combined with the fluid retention index formula and a local overheating risk assessment model, determines the time point at which the static bubble accumulation area faces local overheating risk. If the static bubble accumulation area with severe fluid retention does not face local overheating risk, an infrared thermal imager is used to monitor the temperature and temperature rise rate in real time. A long short-term memory network is used to train the model, predicting the temperature data for the static bubble accumulation area within a preset time period. Based on the temperature data for the preset time period and the local overheating risk assessment model, the time point at which the static bubble accumulation area faces local overheating risk is determined. Based on the predicted time point at which the static bubble accumulation area faces local overheating risk, maintenance measures to remove bubbles from the static bubble accumulation area are implemented in advance.

[0021] Furthermore, the step of evaluating and optimizing the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model based on the local overheating risk data during the implementation of maintenance measures for the static bubble accumulation area includes:

[0022] Based on the local overheating risk data recorded during the implementation of maintenance measures for the static accumulation of bubbles, including bubble size, fluid retention degree, and temperature change data, the accuracy of the models and formulas is evaluated by comparing the actual local overheating risk data with the prediction results of the bubble size prediction model, the fluid retention degree index formula, and the local overheating risk assessment model. If the prediction results of the models and formulas are lower than the preset accuracy threshold, the models and formulas are optimized until the prediction accuracy of the models and formulas reaches the preset requirements. The optimized bubble size prediction model, fluid retention degree index formula, and local overheating risk assessment model are then implemented. The optimization process includes adjusting the parameters of the bubble size prediction model, improving the calculation method of the fluid retention degree index, or redesigning the input features of the local overheating risk assessment model by adding new variables or adjusting the weights of existing features.

[0023] A second aspect of the present invention provides an intelligent management system for a distributed liquid-cooled energy storage system, mainly comprising:

[0024] The bubble acoustic wave reflection signal identification module is used to emit ultrasonic beams through an ultrasonic sensor and receive acoustic wave reflection signals through a directional array sensor. It uses a short-time Fourier transform algorithm to extract the spectral characteristics of the acoustic wave reflection signals and identify acoustic wave reflection signals whose acoustic wave reflection source type is a bubble.

[0025] The bubble static aggregation region identification module is used to acquire the spectral characteristics of the acoustic wave reflection signal of the acoustic wave reflection source type bubble, predict the bubble size, determine the position of the bubble by time difference estimation, and determine the bubble static aggregation region based on the position of the bubble and the distance between the bubbles.

[0026] The fluid retention assessment module is used to determine the bubble aggregation density in the static bubble aggregation area based on the number and size of bubbles in the bubble aggregation area, and to calculate the fluid retention index in the bubble aggregation area to determine the fluid retention degree in the bubble aggregation area.

[0027] The local overheating risk assessment module is used to determine whether there is a local overheating risk in the static bubble accumulation area with severe fluid stagnation based on the load data of the energy storage system and the location and temperature data of the area where the fluid stagnation is severe, and to formulate maintenance measures for the local overheating risk.

[0028] The local overheating risk time point prediction module is used to predict the temperature, bubble size and data of the static bubble accumulation area within a preset time period based on the temperature data, bubble size and number of the static bubble accumulation area where there is no local overheating risk, and to determine the time point when the static bubble accumulation area has a local overheating risk.

[0029] The model evaluation and optimization module is used to evaluate and optimize the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model based on the local overheating risk data when implementing maintenance measures for the overheating risk of the static bubble accumulation area.

[0030] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0031] This invention provides an intelligent management method and system for distributed liquid-cooled energy storage systems. By combining ultrasonic sensing technology with advanced data processing methods, this invention can accurately monitor bubble accumulation in liquid-cooled energy storage systems and achieve real-time identification and location of static bubble accumulation areas. By extracting the spectral characteristics of the bubble acoustic wave reflection signal using a short-time Fourier transform algorithm, this invention can accurately predict the size, number, and location of bubbles, providing reliable data support for the dynamic management of the liquid-cooled system. Through the calculation of the fluid retention index and local overheating risk assessment, this invention can promptly identify and address overheating problems caused by bubble accumulation, thereby avoiding decreased cooling efficiency or equipment failure. This invention not only improves the accuracy of bubble monitoring and fluid management but also predicts bubble change trends based on real-time data and intervenes before potential overheating risks occur, ensuring the stable operation of the energy storage system under high-load conditions. The intelligent management method and system for distributed liquid-cooled energy storage systems provided by this invention significantly improve the accuracy of bubble monitoring, the timeliness of overheat risk assessment, and the optimization efficiency of cooling effect in liquid-cooled energy storage systems. This provides comprehensive and reliable protection for the safe and stable operation of liquid-cooled energy storage systems, extends the service life of equipment, and reduces maintenance costs. Attached Figure Description

[0032] Figure 1 This is a flowchart of a big data-based energy storage battery fault monitoring and management method according to the present invention;

[0033] Figure 2 This is a schematic diagram of a big data-based energy storage battery fault monitoring and management method according to the present invention;

[0034] Figure 3 This is a schematic diagram of a big data-based energy storage battery fault monitoring and management system according to the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0036] like Figure 1-2 This embodiment of an intelligent management method for a distributed liquid-cooled energy storage system may specifically include:

[0037] Step S101: An ultrasonic beam is emitted through an ultrasonic sensor and a sound wave reflection signal is received through a directional array sensor. The spectral characteristics of the sound wave reflection signal are extracted using a short-time Fourier transform algorithm to identify the sound wave reflection signal as a bubble.

[0038] An ultrasonic beam is emitted into the coolant of a distributed liquid-cooled energy storage system using an ultrasonic sensor, and the reflected sound waves are received by a directional array sensor. Based on the received reflected sound waves, an independent component analysis algorithm is used to separate the reflected signals from different sound wave sources. A short-time Fourier transform algorithm is used to convert the time-domain audio signal of the reflected sound waves into frequency-domain data, extracting the spectral features of the reflected sound waves and storing them in a sound wave reflection monitoring database. The spectral features include frequency, peak frequency, and amplitude. Using the sound wave reflection monitoring database, the spectral features of the reflected sound waves from different sources are obtained, and the source type is labeled as whether it is a bubble. A support vector machine algorithm is used to train a model to construct a bubble recognition model, which identifies the reflected sound waves from sources classified as bubbles.

[0039] For example, in a distributed liquid-cooled energy storage system, to accurately monitor air bubbles in the coolant, an ultrasonic sensor emits an ultrasonic beam into the coolant, and a directional array sensor receives the reflected signals of these sound waves. By analyzing the received sound wave reflection signals, different types of sound wave reflection sources can be identified, including but not limited to air bubbles and boundaries of various parts of the energy storage system. The received raw sound wave signal is processed using an independent component analysis (ICA) algorithm to separate different sound wave reflection source signals. When the beam emitted by the ultrasonic sensor reflects off an air bubble in the coolant, the ICA algorithm separates the reflection signal from the air bubble and distinguishes it from other noise or fluid reflection signals. The time-domain audio signal is processed using a short-time Fourier transform (SFT) algorithm to convert it into frequency-domain data. Relevant features are extracted from the frequency distribution of the signal, resulting in a spectrum showing a peak frequency of 25 kHz and an amplitude of approximately 10 dB. This indicates that the signal source may be an air bubble. The spectral characteristics, such as frequency, peak frequency, and amplitude, are stored in the sound wave reflection monitoring database. Based on the accumulated spectral feature data in the database, a support vector machine algorithm is used for training. The model is then optimized using labeled bubble signal samples to construct a bubble recognition model. After training, the model can determine whether the sound wave reflection source is a bubble by using the spectral features of the new sound wave reflection signal. For example, in actual operation, if a new sound wave signal with a peak frequency of 25.5kHz and an amplitude of 9dB is received, the trained bubble recognition model can correctly determine that the signal comes from a bubble.

[0040] Step S102: Obtain the spectral characteristics of the acoustic wave reflection signal of the bubble as the acoustic wave reflection source type, predict the bubble size, determine the bubble position by time difference estimation, and determine the static bubble aggregation area based on the bubble position and the distance between bubbles.

[0041] Based on the real-time received acoustic wave reflection signals, a bubble recognition model is used to identify acoustic wave reflection signals whose source type is bubbles. The number of acoustic wave reflection signals with bubble source type and their corresponding spectral characteristics are obtained. A random forest regression algorithm is used to train the model, constructing a bubble size prediction model to predict bubble size. Based on the echo time of the bubble acoustic wave reflection signals acquired by the directional array sensor, the distance of the bubble relative to the sensor array is calculated through time difference estimation to determine the bubble's position. The positional changes of the bubbles are continuously monitored, and the distances between bubbles are calculated based on their positions. Static bubble aggregation areas are determined based on the changes in bubble position and the distances between bubbles.

[0042] For example, in a distributed liquid-cooled energy storage system, ultrasonic sensors emit sound waves and receive their reflected signals to monitor air bubbles in the coolant. If, at a certain moment, the received sound wave reflection signal is analyzed, a pre-trained bubble recognition model is used to classify the reflection signal, indicating that the reflection source is a bubble. Using historical data from a database and the spectral characteristics of the current sound wave reflection signal, a random forest regression algorithm is used to train a bubble size prediction model to predict the bubble size. The trained model predicts the bubble size to be 2.5 mm. Based on the echo time obtained from the directional array sensor, the distance of the bubble relative to the sensor array is calculated. If the propagation speed of the ultrasonic signal emitted by the sensor array is 1500 m / s, and the time difference of the echo signal is 0.005 seconds, the time difference estimation yields a distance of 7.5 meters between the bubble and the sensor array. As the system continues to operate, it monitors the positional changes of the bubbles in real time. If, at the next moment, it receives a sound wave reflection signal from another bubble, and estimates the distance of that bubble to be 8 meters based on the echo time, while the distance between the bubbles is 1.5 meters, it can determine the positional changes of the bubbles by calculating these distances. Based on this positional data, it can identify whether the bubbles are in the same cluster area. For example, if the positional changes of two bubbles are small and their distance is close, and their predicted sizes are consistent with historical data, it can be determined that they have formed a static bubble cluster area. Bubbles in this area may cause a decrease in the efficiency of the cooling system, especially at high temperatures.

[0043] Step S103: Based on the number and size of bubbles in the static bubble aggregation area, determine the bubble aggregation density in the static bubble aggregation area, calculate the fluid retention index in the bubble aggregation area, and judge the fluid retention degree in the bubble aggregation area.

[0044] To obtain the number and size of bubbles within the static bubble aggregation region, use the bubble aggregation density factor formula. Calculate the bubble aggregation density ρ in the static bubble aggregation region. b ′, where N b It is the number of bubbles, D b It is the average size of the bubbles, I avg This is the average intensity of the sound wave reflection signal within the static aggregation region of the bubbles. The bubble interaction factor formula is used based on the bubble size and the distance between bubbles. Determine the bubble interaction factor α b ′, where d b It is the minimum distance between bubbles, α b The impact of reaction bubble aggregation on flow is assessed. Based on bubble aggregation density and bubble interaction factor, the fluid retention index formula is used. A fluid retention index R is determined within the bubble aggregation region, where β and γ are adjustment factors obtained through fitting historical data, n is the index of the effect of bubble aggregation density on retention, used to control the rate of increase in retention as density increases, and ρ0′ is the baseline value of bubble aggregation density under normal operating conditions. Based on the fluid retention index within the bubble aggregation region and a preset index threshold, the degree of fluid retention within the bubble aggregation region is determined, categorized as severe, moderate, and moderate.

[0045] For example, a bubble aggregation region in a liquid cooling system is being monitored, and relevant reflected signal data is obtained from an ultrasonic sensor. The number and average size of the bubbles in this region are acquired. If there are 50 bubbles, each with an average diameter of 2 mm, and the average intensity of the reflected sound signal in this region is 10 dB, the bubble aggregation density factor formula is used. Calculate the bubble aggregation density factor ρ b ′, where N b It is the number of bubbles, D b It is the average size of the bubbles, I avg The average intensity of the sound wave reflection signal within the static bubble aggregation region is used to calculate the bubble aggregation density factor ρ. b The value is 10. Based on the bubble size and the minimum distance between bubbles, if the minimum distance between bubbles is 0.5 mm, the bubble interaction factor formula is used. Determine the bubble interaction factor α b ′, where d b It is the minimum distance between bubbles, α b The extent to which reaction bubble aggregation affects the flow was calculated, and the bubble interaction factor α was obtained. b The value is 0.0588. Based on the bubble aggregation density factor and bubble interaction factor, the fluid retention index formula is used. The fluid retention index R is calculated, where β and γ are adjustment factors obtained by fitting historical data, n is the index of the effect of bubble aggregation density on retention, used to control the rate of increase in retention as density increases, and ρ0′ is the baseline value of bubble aggregation density under normal operating conditions. If the adjustment factor β = 1.2, the baseline value of bubble aggregation density ρ0′ = 5, n = 1.5, and γ = 0.1 obtained from fitting historical data, the retention index is 3.38. If the preset threshold for the normal retention index is 3.5, the threshold for the moderate retention index is 4.0, and the threshold for the severe retention index is 4.5, and the calculated retention index is 3.38, which is lower than the preset threshold for the normal retention index, it can be determined that the fluid retention in this area is normal, and no overheating risk maintenance measures are required.

[0046] Step S104: Based on the load data of the energy storage system and the location and temperature data of the static bubble accumulation area with severe fluid stagnation, determine whether there is a risk of local overheating in the static bubble accumulation area with severe fluid stagnation, and formulate maintenance measures for the local overheating risk.

[0047] The location of static bubble accumulation areas is determined, including pipe bends, joints, changes in location, and heat dissipation components. Load data of the energy storage system is acquired through the battery management system, and temperature data of severely stagnant static bubble accumulation areas is obtained using an infrared thermal imager. Load data includes current, voltage, and power. Based on the energy storage system's load data and the location and temperature data of severely stagnant static bubble accumulation areas, a recurrent neural network is used to train a model to construct a local overheating risk assessment model. Based on the real-time acquired load data of the energy storage system and the location and temperature data of severely stagnant static bubble accumulation areas, the local overheating risk assessment model is used to determine whether there is a risk of local overheating in these areas. If a risk of local overheating exists, maintenance measures are formulated and implemented, including but not limited to adjusting the flow rate of the coolant within the energy storage system and adjusting the pressure of the liquid cooling system to remove bubbles from the static bubble accumulation areas.

[0048] For example, in a distributed liquid-cooled energy storage system, ultrasonic sensors and infrared thermal imagers detected a severe static bubble accumulation area in the coolant. This area was located at a bend in the pipe, and temperature data indicated that the temperature in this area reached as high as 60°C. The significant fluid stagnation indicated that coolant flow was impeded, and the bubble accumulation led to heat accumulation. Simultaneously, load data from the battery management system showed that at that moment, the energy storage system's current was 20A, voltage was 400V, and power was 8kW, meaning the system was under high load, potentially exacerbating the overheating risk in the bubble accumulation area. A recurrent neural network was used to train a model to construct a local overheating risk assessment model. The model input included real-time energy storage system load data, such as current, voltage, and power, as well as the temperature and location data of the severely stagnant static bubble accumulation area. When new load data was acquired in real-time, such as a current of 25A, voltage of 420V, and power of 10.5kW, and the temperature sensor detected that the temperature in the bubble accumulation area had risen again to 65°C, the trained local overheating risk assessment model determined that there was a local overheating risk in this area. Based on the model's predictions, the system triggered an early warning and initiated maintenance measures to address the localized overheating risk. To mitigate the overheating problem, firstly, the coolant flow rate was increased by 10% to promote bubble movement and heat dissipation. Secondly, the pressure of the liquid cooling system was adjusted to an appropriate level to reduce bubble aggregation and further stagnation. Through these measures, the temperature in the static bubble aggregation area gradually decreased, coolant flow was restored, and the risk of localized overheating was effectively mitigated.

[0049] Step S105: Based on the temperature data, bubble size, and number of the static bubble accumulation area where there is no risk of local overheating, predict the temperature, bubble size, and data of the static bubble accumulation area within a preset time period in the future, and determine the time point when there is a risk of local overheating in the static bubble accumulation area.

[0050] If the fluid retention level in the static bubble accumulation area is moderate or average, the size and number of bubbles in this area are continuously acquired. A long short-term memory (LSTM) network is used to train a model to predict the bubble size and number within a preset future time period. This prediction, combined with the fluid retention index formula and a local overheating risk assessment model, determines the time point at which the static bubble accumulation area faces local overheating risk. If the static bubble accumulation area with severe fluid retention does not face local overheating risk, an infrared thermal imager is used to monitor the temperature and temperature rise rate in real time. An LTM network is then used to train a model to predict the temperature data of the static bubble accumulation area within a preset future time period. Based on the temperature data within this preset future time period, and combined with the local overheating risk assessment model, the time point at which the static bubble accumulation area faces local overheating risk is determined. Based on the predicted time point at which the static bubble accumulation area faces local overheating risk, preventative measures are implemented to remove bubbles from the static bubble accumulation area.

[0051] For example, in a liquid-cooled energy storage system, the fluid retention level in a static bubble accumulation area near the radiator is determined to be moderate. Using an ultrasonic sensor, we obtain data on the size and number of bubbles in this area. Currently, there are 100 bubbles, each with an average diameter of 1.8 mm. Using a Long Short-Term Memory (LSTM) network, we train a model to predict changes in the number and size of bubbles over the next 10 minutes. The model predicts that the number of bubbles will increase to 120 and the average diameter will increase to 2.1 mm within the next 10 minutes. This means that bubble accumulation may gradually intensify, affecting the cooling effect. Based on the bubble density and predicted bubble size, combined with the fluid retention index formula and a local overheating risk assessment model, we further predict the future local overheating risk of this static bubble accumulation area. If the predicted fluid retention index for this static bubble accumulation area is 4.8 within the next five minutes, exceeding the preset severe retention index threshold of 4.5, and the local overheating risk assessment model assesses that the static bubble accumulation area may experience local overheating within the next five minutes, then... However, if, under certain circumstances, the fluid stagnation in the static bubble accumulation area is severe, but the risk of local overheating is not detected by the model, an infrared thermal imager is used to monitor the temperature of the area in real time. Using the real-time temperature data, a long short-term memory network model is used again to predict the temperature change of the static bubble accumulation area in the next 10 minutes. If the model predicts that the temperature of the area will further rise to 70°C in the next 10 minutes, combined with the local overheating risk assessment model, it is determined that the area will experience local overheating risk in the next 8 minutes. It is recommended to implement cooling and bubble removal measures in advance to avoid more serious consequences caused by overheating.

[0052] Step S106: Based on the local overheating risk data when implementing maintenance measures for the overheating risk of the static bubble accumulation area, evaluate and optimize the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model.

[0053] Based on the local overheating risk data recorded during the implementation of maintenance measures for the static accumulation of bubbles, including bubble size, fluid retention degree, and temperature change data, the accuracy of the models and formulas is evaluated by comparing the actual local overheating risk data with the prediction results of the bubble size prediction model, the fluid retention degree index formula, and the local overheating risk assessment model. If the prediction results of the models and formulas are lower than the preset accuracy threshold, the models and formulas are optimized until the prediction accuracy of the models and formulas reaches the preset requirements. The optimized bubble size prediction model, fluid retention degree index formula, and local overheating risk assessment model are then implemented. The optimization process includes adjusting the parameters of the bubble size prediction model, improving the calculation method of the fluid retention degree index, or redesigning the input features of the local overheating risk assessment model by adding new variables or adjusting the weights of existing features.

[0054] For example, in a liquid-cooled energy storage system, during the implementation of overheating risk maintenance measures for a static bubble accumulation area, actual overheating data for that area was recorded. Monitoring with an infrared thermal imager showed that the temperature in the static bubble accumulation area increased from 65°C to 72°C, while the fluid retention level was moderate before the maintenance measures. The bubbles were 2.1 mm in size and numbered 125. These actual data were compared with data previously predicted using a bubble size prediction model and a local overheating risk assessment model. According to the model prediction, the bubble size should be 2.35 mm and the number should be 115, and the local overheating risk assessment model predicted the temperature in that area should reach 68°C, showing a certain difference. By comparing the actual data with the predicted results, it was found that there was a certain error between the predicted and actual values, especially in terms of bubble size and temperature rise; the model's accuracy failed to reach the preset threshold of 90%. At this point, based on the preset standards, it was determined that the current model's prediction accuracy was low, and therefore it was decided to optimize the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model. The optimization process included improving the model's accuracy in predicting bubble size by adjusting hyperparameters, such as increasing the size of the training dataset or introducing more features like the rate of temperature change. The calculation method for the fluid retention index formula was also improved by incorporating more influencing factors, such as coolant flow rate variations and pipe wall roughness, to enhance the formula's accuracy in estimating retention. The input features of the local overheating risk assessment model were redesigned, incorporating new variables such as bubble distribution and the overall efficiency of the liquid cooling system. The weights of existing features were also adjusted to enable the model to more accurately predict overheating risks. After these optimizations, the new model and formula were applied to the actual system. In subsequent tests, the optimized model's predictions showed a significant reduction in error compared to the actual data. The predicted bubble size was 2.4 mm with 123 bubbles, and the predicted temperature was 70°C, closely matching the actual monitored data. Furthermore, the model's prediction accuracy reached 92%, successfully meeting the preset accuracy requirements.

[0055] like Figure 3 This embodiment of an intelligent management system for a distributed liquid-cooled energy storage system may specifically include:

[0056] The bubble acoustic wave reflection signal identification module is used to emit ultrasonic beams through an ultrasonic sensor and receive acoustic wave reflection signals through a directional array sensor. It uses a short-time Fourier transform algorithm to extract the spectral characteristics of the acoustic wave reflection signals and identify acoustic wave reflection signals whose source type is a bubble.

[0057] The bubble static aggregation region identification module is used to acquire the spectral characteristics of the acoustic wave reflection signal of the acoustic wave reflection source type bubble, predict the bubble size, determine the position of the bubble by time difference estimation, and determine the static aggregation region of the bubble based on the position of the bubble and the distance between the bubbles.

[0058] The fluid retention assessment module is used to determine the bubble aggregation density in the static bubble aggregation area based on the number and size of bubbles in the bubble aggregation area, and to calculate the fluid retention index in the bubble aggregation area to determine the fluid retention degree in the bubble aggregation area.

[0059] The local overheating risk assessment module is used to determine whether there is a local overheating risk in the static bubble accumulation area with severe fluid stagnation based on the load data of the energy storage system and the location and temperature data of the area. It also formulates maintenance measures to address the local overheating risk.

[0060] The local overheating risk prediction module is used to predict the temperature, bubble size, and data of a static bubble accumulation area within a preset time period based on the temperature data, bubble size, and number of the static bubble accumulation area where there is no local overheating risk, and to determine the time point when the static bubble accumulation area has a local overheating risk.

[0061] The model evaluation and optimization module is used to evaluate and optimize the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model based on the local overheating risk data when implementing maintenance measures for the overheating risk of the static bubble accumulation area.

[0062] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the concept of this application. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for intelligent management of a distributed liquid-cooled energy storage system, characterized in that, The method includes: An ultrasonic beam is emitted by an ultrasonic sensor and the reflected sound wave signal is received by a directional array sensor. The spectral characteristics of the reflected sound wave signal are extracted using a short-time Fourier transform algorithm to identify the sound wave reflection source as a bubble. The spectral characteristics of the acoustic wave reflection signal of the bubble as the acoustic wave reflection source are obtained, the bubble size is predicted, the bubble position is determined by time difference estimation, and the static bubble aggregation region is determined based on the bubble position and the distance between bubbles. Based on the number and size of bubbles in the static bubble aggregation region, the bubble aggregation density in the static bubble aggregation region is determined, and the fluid retention index in the bubble aggregation region is calculated to determine the fluid retention degree in the bubble aggregation region. Based on the load data of the energy storage system and the location and temperature data of the static bubble accumulation area with severe fluid stagnation, determine whether there is a risk of local overheating in the static bubble accumulation area with severe fluid stagnation, and formulate maintenance measures for local overheating risk. Based on the temperature data, bubble size, and number of the static bubble accumulation area where there is no risk of local overheating, predict the temperature, bubble size, and data of the static bubble accumulation area within a preset time period in the future, and determine the time point when there is a risk of local overheating in the static bubble accumulation area. Based on the local overheating risk data when implementing maintenance measures for the overheating risk of static bubble accumulation areas, the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model is evaluated and optimized. The process of determining the bubble aggregation density within the static bubble aggregation region based on the number and size of bubbles, and calculating the fluid retention index within the bubble aggregation region to determine the fluid retention degree within the bubble aggregation region, includes: To obtain the number and size of bubbles within the static bubble aggregation region, use the bubble aggregation density factor formula. Calculate the bubble aggregation density in the static bubble aggregation region. ,in, It refers to the number of bubbles. It is the average size of the bubbles. It is the average intensity of the sound wave reflection signal within the static aggregation region of the bubbles; based on the bubble size and the distance between the bubbles, the bubble interaction factor formula is used. Determine the bubble interaction factor ,in, It is the minimum distance between bubbles. The impact of reaction bubble aggregation on flow was assessed; based on bubble aggregation density and bubble interaction factor, a fluid retention index formula was used. Determine the fluid retention index R within the bubble aggregation region, where, and This is a regulating factor, obtained through fitting historical data. 'n' is an exponential function of bubble aggregation density on retention, used to control the rate at which retention increases with density. It is the baseline value of bubble aggregation density under normal operating conditions; based on the fluid retention index and preset index threshold in the bubble aggregation area, the fluid retention degree in the bubble aggregation area is determined, and the fluid retention degree includes severe, moderate and ordinary.

2. The method according to claim 1, wherein, The process involves emitting an ultrasonic beam via an ultrasonic sensor, receiving the reflected sound wave signal via a directional array sensor, extracting the spectral characteristics of the reflected sound wave signal using a short-time Fourier transform algorithm, and identifying the reflected sound wave signal as a bubble as the source of the reflected sound wave. This includes: An ultrasonic beam is emitted into the coolant of a distributed liquid-cooled energy storage system using an ultrasonic sensor, and the reflected sound wave signal is received by a directional array sensor. Based on the received reflected sound wave signal, an independent component analysis algorithm is used to separate the reflected sound wave signal from different sound wave sources. A short-time Fourier transform algorithm is used to convert the time-domain audio signal of the reflected sound wave signal into frequency-domain data, extract the spectral features of the reflected sound wave signal, and save them to a sound wave reflection monitoring database. The spectral features include frequency, peak frequency, and amplitude. Through the sound wave reflection monitoring database, the spectral features of the reflected sound wave signals from different sound wave sources are obtained, and the type of sound wave reflection source is labeled as a bubble. A support vector machine algorithm is used to train the model and build a bubble recognition model to identify the reflected sound wave signal of a bubble source.

3. The method according to claim 1, wherein, The process of acquiring the spectral characteristics of the acoustic wave reflection signal of a bubble as the acoustic wave reflection source, predicting the bubble size, determining the bubble position through time difference estimation, and determining the static bubble aggregation region based on the bubble position and the distance between bubbles includes: Based on the real-time received acoustic wave reflection signals, a bubble recognition model is used to identify acoustic wave reflection signals whose source type is bubbles. The number of acoustic wave reflection signals whose source type is bubbles and the spectral characteristics of the corresponding acoustic wave reflection signals are obtained. A random forest regression algorithm is used to train the model and construct a bubble size prediction model to predict the bubble size. Based on the echo time of the bubble acoustic wave reflection signals obtained by the directional array sensor, the distance of the bubble relative to the sensor array is calculated through time difference estimation to determine the position of the bubble. The position changes of the bubbles are continuously monitored, and the distance between the bubbles is calculated based on the position changes of the bubbles. Based on the position changes of the bubbles and the distance between the bubbles, the static bubble aggregation area is determined.

4. The method according to claim 1, wherein, Based on the load data of the energy storage system and the location and temperature data of the static bubble accumulation area with severe fluid stagnation, it is determined whether there is a risk of local overheating in the static bubble accumulation area with severe fluid stagnation, and maintenance measures for local overheating risk are formulated, including: The system acquires the location of static bubble accumulation areas, including pipe bends, joints, changes in location, and heat dissipation components. It obtains load data from the energy storage system via the battery management system and temperature data of severely stagnant static bubble accumulation areas using an infrared thermal imager. Load data includes current, voltage, and power. Based on the energy storage system's load data and the location and temperature data of severely stagnant static bubble accumulation areas, a recurrent neural network is used to train a model to construct a local overheating risk assessment model. Using the real-time acquired load data and the location and temperature data of severely stagnant static bubble accumulation areas, the local overheating risk assessment model is used to determine whether there is a risk of local overheating in these areas. If a local overheating risk is identified, maintenance measures are developed and implemented, including but not limited to adjusting the flow rate of the coolant within the energy storage system and adjusting the pressure of the liquid cooling system to remove bubbles from the static bubble accumulation areas.

5. The method according to claim 1, wherein, The process of predicting the temperature, bubble size, and number of statically accumulated bubbles within a preset time period based on temperature data, bubble size, and number of areas without local overheating risk, and determining the time points where local overheating risk exists in the statically accumulated bubbles area, includes: If the fluid retention level in the static bubble accumulation area is moderate or normal, the size and number of bubbles in the static bubble accumulation area are continuously acquired, and a long short-term memory network is used for model training to predict the size and number of bubbles in the future within a preset time period. The fluid retention level index formula and the local overheating risk assessment model are combined to determine the time point when the static bubble accumulation area has a local overheating risk. If there is no risk of local overheating in the static bubble accumulation area where fluid stagnation is severe, the temperature of the static bubble accumulation area is monitored in real time using an infrared thermal imager to obtain the temperature and temperature rise rate of the static bubble accumulation area. A long short-term memory network is used to train a model to predict the temperature data of the static bubble accumulation area within a preset time period. Based on the temperature data of the static bubble accumulation area within the preset time period, combined with the local overheating risk assessment model, the time point when the static bubble accumulation area has a local overheating risk is determined. Based on the predicted time point when the static bubble accumulation area has a local overheating risk, local overheating risk maintenance measures are implemented in advance to remove bubbles from the static bubble accumulation area.

6. The method according to claim 1, wherein, The process of evaluating and optimizing the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model based on local overheating risk data from the implementation of maintenance measures for static bubble accumulation areas includes: Based on the local overheating risk data recorded during the implementation of maintenance measures for the static accumulation of bubbles, including bubble size, fluid retention degree, and temperature change data, the accuracy of the models and formulas is evaluated by comparing the actual local overheating risk data with the prediction results of the bubble size prediction model, the fluid retention degree index formula, and the local overheating risk assessment model. If the prediction results of the models and formulas are lower than the preset accuracy threshold, the models and formulas are optimized until the prediction accuracy of the models and formulas reaches the preset requirements. The optimized bubble size prediction model, fluid retention degree index formula, and local overheating risk assessment model are then implemented. The optimization process includes adjusting the parameters of the bubble size prediction model, improving the calculation method of the fluid retention degree index, or redesigning the input features of the local overheating risk assessment model by adding new variables or adjusting the weights of existing features.

7. An intelligent management system for a distributed liquid-cooled energy storage system, implemented based on the intelligent management method for a distributed liquid-cooled energy storage system as described in any one of claims 1-6, characterized in that, The system includes the following modules: The bubble acoustic wave reflection signal identification module is used to emit ultrasonic beams through an ultrasonic sensor and receive acoustic wave reflection signals through a directional array sensor. It uses a short-time Fourier transform algorithm to extract the spectral characteristics of the acoustic wave reflection signals and identify acoustic wave reflection signals whose acoustic wave reflection source type is a bubble. The bubble static aggregation region identification module is used to acquire the spectral characteristics of the acoustic wave reflection signal of the acoustic wave reflection source type bubble, predict the bubble size, determine the position of the bubble by time difference estimation, and determine the bubble static aggregation region based on the position of the bubble and the distance between the bubbles. The fluid retention assessment module is used to determine the bubble aggregation density in the static bubble aggregation area based on the number and size of bubbles in the bubble aggregation area, and to calculate the fluid retention index in the bubble aggregation area to determine the fluid retention degree in the bubble aggregation area. The local overheating risk assessment module is used to determine whether there is a local overheating risk in the static bubble accumulation area with severe fluid stagnation based on the load data of the energy storage system and the location and temperature data of the area where the fluid stagnation is severe, and to formulate maintenance measures for the local overheating risk. The local overheating risk time point prediction module is used to predict the temperature, bubble size and data of the static bubble accumulation area within a preset time period based on the temperature data, bubble size and number of the static bubble accumulation area where there is no local overheating risk, and to determine the time point when the static bubble accumulation area has a local overheating risk. The model evaluation and optimization module is used to evaluate and optimize the accuracy of the bubble size prediction model, the fluid retention index formula, and the local overheating risk assessment model based on the local overheating risk data when implementing maintenance measures for the overheating risk of the static bubble accumulation area.

Images