Battery cooling liquid intelligent deployment and component optimization method based on multi-source data fusion

Abstract

The application discloses a battery cooling liquid intelligent deployment and component optimization method based on multi-source data fusion, and belongs to the field of battery thermal management. The method collects battery temperature, current, voltage and environmental temperature data through a multi-source data fusion module, generates a thermal load evaluation index by using a weighted fusion algorithm, and divides a working condition grade; candidate proportion ranges are screened from a three-dimensional parameter performance database covering nano-particle concentration, base liquid proportion and additive content according to the working condition grade; a multi-objective optimization algorithm is called to solve the component proportion that is optimal in comprehensive cooling efficiency, pumping energy consumption and cost, with the candidate ranges as constraints; finally, each raw liquid in a multi-chamber liquid storage system is dynamically mixed according to the optimal proportion through a high-precision metering pump and an electric proportional valve joint control system. The application realizes closed-loop linkage of working condition sensing and component optimization, can dynamically adjust the cooling liquid formula in real time, and takes into account both heat dissipation performance and energy consumption economy.

Description

Technical Field

[0001] This application belongs to the field of battery intelligent management technology, specifically relating to a method for intelligent allocation and component optimization of battery coolant based on multi-source data fusion. Background Technology

[0002] In battery thermal management systems, the thermal conductivity and viscosity ratio of the coolant directly determine the balance between heat dissipation efficiency and pumping energy consumption. The core challenge we face is how to dynamically adjust the coolant composition ratio according to the real-time thermal load of the battery to achieve optimal cooling performance.

[0003] Common battery coolant optimization methods typically employ fixed-formulation nanofluids or multi-component premixing schemes based on expert experience. However, neither of these methods can dynamically adjust the composition according to real-time changes in battery operating conditions. Fixed-formulation schemes ignore the significant differences in battery thermal load under different operating conditions, resulting in a discrepancy between coolant performance and actual heat dissipation requirements that is difficult to bridge. On the other hand, multi-component schemes, lacking data fusion mechanisms, cannot accurately perceive the real-time thermal load status of the battery, and their component ratio adjustments lack the necessary operating condition basis, making it difficult to achieve the optimal balance between cooling efficiency and energy consumption control.

[0004] Establishing a closed-loop linkage mechanism between operating condition perception and component optimization, and dynamically calculating and adjusting the coolant composition ratio based on the real-time heat load of the battery, has become a key challenge that urgently needs to be addressed in the field of battery thermal management. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion, which can effectively solve the problems mentioned in the background technology.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: This invention provides a method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion, comprising: The multi-source data fusion module collects multi-dimensional sensor data of battery temperature, current, voltage and ambient temperature. The weighted fusion algorithm is used to preprocess and deeply fuse the multi-dimensional sensor data to generate a battery thermal load assessment index that reflects the real-time thermal load status of the battery. The current thermal load condition is divided into different levels according to a preset threshold. Based on the level of the current heat load condition, candidate component ratio ranges that match the current heat load condition are selected from a pre-built coolant component performance database; wherein, the coolant component performance database is a three-dimensional parameter mapping table covering three dimensions: nanoparticle concentration, base fluid ratio, and additive content. Using the candidate group allocation ratio range as a search space constraint, the component optimization algorithm is invoked to perform multi-objective optimization calculation. The optimization objectives of the multi-objective optimization calculation include at least improving cooling efficiency, reducing pumping energy consumption, and cost constraints. The optimal group allocation ratio under the current operating conditions is solved and output. The system uses a high-precision metering pump and an electric proportional valve to physically mix and dynamically adjust the base liquid stock solution, nanoparticle dispersion and functional additive stock solution stored in the multi-chamber coolant storage system according to the optimal composition ratio.

[0007] Furthermore, the generation of the battery thermal load assessment index, which reflects the real-time thermal load state of the battery, specifically includes: The timing alignment module resamples the raw data collected by each data acquisition channel under the global clock drive within the timing alignment module, according to the nearest timestamp alignment principle, in order to eliminate timing misalignment caused by sampling frequency differences. The multidimensional data after time-series synchronization is processed by a weighted fusion algorithm module, which includes a data preprocessing layer, a dynamic weight calculation layer, and a fusion output layer. In the dynamic weight calculation layer, the current weight coefficient for each data acquisition channel is dynamically calculated based on the confidence level and historical accuracy of the data from each data acquisition channel. The confidence level evaluation indicators include the validity of the sensor self-test status signal, whether the rate of change of the data acquisition channel data is within a reasonable physical range, and the correlation analysis results between the data acquisition channel data and other data acquisition channels. When the data from a certain data acquisition channel is determined to be abnormal, the confidence level of the abnormal data acquisition channel is set to 0.2, causing the final weight of the abnormal data acquisition channel after normalization calculation to drop below 20% of the normal value. In the fusion output layer, the collected data values ​​of each data acquisition channel are multiplied by their corresponding weight coefficients and then summed to generate the fused battery thermal load evaluation index.

[0008] Furthermore, the steps for establishing the coolant component performance database include: A three-dimensional parameter mapping table is established. The first dimension of the three-dimensional parameter mapping table is the nanoparticle concentration, covering multiple concentration gradient nodes from 0.1% to 2.0%. The second dimension of the three-dimensional parameter mapping table is the base liquid ratio, which is defined as the volume percentage combination of water and ethylene glycol, covering multiple ratio gradient nodes from pure water to pure ethylene glycol. The third dimension of the three-dimensional parameter mapping table is the additive content, covering multiple content gradient nodes from 0.01% to 1.0%. The gradient nodes of the three dimensions are cross-combined to form a three-dimensional parameter space. For each group allocation scheme in the three-dimensional parameter space, a corresponding performance parameter field is configured. The performance parameter field includes at least thermal conductivity, dynamic viscosity, cooling efficiency index, and comprehensive score.

[0009] Furthermore, the invocation of the component optimization algorithm to perform multi-objective optimization calculations and solve for the optimal component allocation ratio under the current operating conditions specifically includes: The component optimization algorithm is solved iteratively using an improved multi-objective particle swarm optimization algorithm or a multi-objective genetic algorithm; The improved multi-objective particle swarm optimization algorithm introduces an adaptive inertial weight to adjust the flight speed of particles during the iteration process. The adaptive inertial weight is reduced when the particles approach the optimal solution and increased when the particles move away from the optimal solution. After obtaining the Pareto optimal solution set, the final implementation scheme is selected from the Pareto optimal solution set as the optimal group allocation ratio according to the preset decision strategy; wherein, the decision strategy adopts the weighted summation method, and the weight coefficient of the weighted summation method is dynamically adjusted according to the level of the current heat load condition. When the heat load is at the high load condition level or the extremely high load condition level, the weight coefficient of the cooling efficiency target is set to the highest value.

[0010] Furthermore, the physical mixing and dynamic adjustment via a combined control system of a high-precision metering pump and an electric proportional valve specifically includes: The high-precision metering pump is installed on the outlet pipe of each storage chamber of the multi-chamber coolant storage system. The response time of the high-precision metering pump is no more than 100 milliseconds, and the metering accuracy of the high-precision metering pump is controlled within ±0.5%. The high-precision metering pump calibrates the deviation between the actual output flow and the set flow in real time through a built-in closed-loop flow feedback control loop. The electric proportional valve is installed on the connecting pipeline between the high-precision metering pump and the mixing pipeline. The control accuracy of the electric proportional valve is controlled within ±0.3%. The electric proportional valve realizes continuous stepless adjustment of the valve opening by receiving pulse width modulation control signal. The high-precision metering pump controls the basic flow output of the base liquid stock solution, nanoparticle dispersion, and functional additive stock solution. The electric proportional valve performs flow fine-tuning and compensation based on the basic flow controlled by the high-precision metering pump. The main control unit coordinates and outputs the control signals of the high-precision metering pump and the electric proportional valve to ensure that the actual flow ratio of the base liquid stock solution, nanoparticle dispersion, and functional additive stock solution remains consistent with the optimal composition ratio.

[0011] Furthermore, the multi-chamber coolant storage system includes at least three independent storage chambers. The first storage chamber is used to store the base liquid stock solution composed of a mixture of water and ethylene glycol. The second storage chamber is used to store a nanoparticle dispersion formed by uniformly dispersing nanoparticles in the base liquid stock solution. The third storage chamber is used to store a functional additive stock solution containing corrosion inhibitor components, scale inhibitor components, and corrosion inhibitor components. Each storage chamber is equipped with an independent capacitive level gauge and an independent platinum resistance temperature sensor to monitor the changes in the liquid level and the temperature changes of the stored stock solution in real time.

[0012] Furthermore, after adjusting the group allocation ratio, a feedback correction mechanism is also included, which specifically includes: Continuously monitor battery temperature change trends and thermal management system energy consumption data, and calculate the cooling efficiency deviation and energy consumption deviation between the actual cooling effect and the optimization target; A model predictive control strategy is adopted, and a cooling effect prediction model is established based on historical operating data. The battery thermal load assessment index, the ambient temperature value collected by the ambient temperature sensor, the coolant flow rate value collected by the coolant flow sensor, and the current component ratio are used as input variables, and the battery temperature rise rate and system energy consumption are used as output variables. The cooling effect prediction model adopts a multiple linear regression model. When either the cooling efficiency deviation or the energy consumption deviation exceeds the preset allowable range, the main control unit automatically triggers a new round of optimization calculation.

[0013] Furthermore, the feedback correction mechanism supports an incremental adjustment strategy. Each time a new round of optimization calculation is triggered, only a small adjustment is made based on the current ratio. Specifically, the adjustment step size for nanoparticle concentration does not exceed ±5% of the current nanoparticle concentration value, the adjustment step size for base liquid ratio does not exceed ±3% of the current base liquid ratio value, and the adjustment step size for additive content does not exceed ±5% of the current additive content value. The specific values ​​of the adjustment step size for nanoparticle concentration, base liquid ratio, and additive content are automatically determined by the component optimization algorithm based on the current cooling efficiency deviation value and energy consumption deviation value.

[0014] Furthermore, the establishment of the coolant component performance database was completed through systematic experimental testing, specifically including: In the first phase of testing, the base solution ratio and additive content were kept constant, and various performance parameters were measured sequentially at different nanoparticle concentrations to form a curve showing the relationship between nanoparticle concentration and performance parameters. In the second phase of testing, while keeping the concentration of nanoparticles and the content of additives constant, various performance parameters were measured sequentially under different base solution ratios to form a curve showing the relationship between the base solution ratio and the performance parameters. In the third stage of testing, while keeping the concentration of nanoparticles and the ratio of base solution constant, various performance parameters were measured sequentially under different additive contents to form a curve showing the relationship between additive content and performance parameters. For the discrete measurement data points obtained from the first stage test, the second stage test, and the third stage test, the cubic spline interpolation method is used to perform curve fitting, and the three fitted curves are extended by Cartesian product in the three-dimensional parameter space to form a complete three-dimensional parameter performance mapping space.

[0015] Furthermore, the step of selecting candidate component allocation ratio ranges that match the current thermal load conditions from a pre-built coolant component performance database specifically includes: When the battery thermal load evaluation index is under low load conditions, the formulation scheme with a low thermal conductivity and low dynamic viscosity is selected as a candidate. When the battery thermal load evaluation index is under high load or extremely high load conditions, the ratio scheme with a medium to high thermal conductivity and a high cooling efficiency index is selected as a candidate.

[0016] In summary, this application includes at least one of the following beneficial technical effects: 1. This invention achieves a closed-loop dynamic linkage between operating condition perception and component optimization. The invention uses a multi-source data fusion module to collect real-time multi-dimensional sensor data such as battery temperature, current, voltage, and ambient temperature. A weighted fusion algorithm is used to generate an evaluation index reflecting the real-time thermal load state of the battery and classify operating condition levels. Based on this, candidate ratio ranges are screened from a three-dimensional parameter performance database, and an optimization algorithm is called to solve for the optimal ratio. Finally, a precision actuator completes the physical mixing. The entire process forms a complete closed loop from state perception, database matching, optimization decision-making to precise execution and feedback correction, enabling the coolant composition to dynamically and adaptively adjust according to real-time changes in battery operating conditions. This effectively overcomes the fundamental deficiency of traditional fixed-formulation solutions that cannot adjust according to changes in thermal load.

[0017] 2. This invention achieves a comprehensive optimization of multiple objectives, including cooling efficiency, pumping energy consumption, and cost constraints. It constructs a three-dimensional parameter performance mapping space encompassing nanoparticle concentration, base fluid ratio, and additive content. An improved multi-objective particle swarm optimization algorithm or multi-objective genetic algorithm is used for iterative solving. This simultaneously incorporates improved cooling efficiency, reduced pumping energy consumption, and cost constraints into the optimization objective function system. Under high load conditions, priority is given to ensuring heat dissipation capacity, while under low load conditions, the focus is on reducing energy consumption. Thus, under different operating scenarios, a comprehensive optimal composition ratio that balances heat dissipation performance, flow economy, and economic cost can be obtained, resolving the contradiction of balancing cooling performance and energy consumption control in traditional methods.

[0018] 3. This invention improves the accuracy of proportioning and the stability of system operation. Through a high-precision metering pump and electric proportional valve combined control system, along with a closed-loop flow feedback control loop, this invention controls the metering accuracy within ±0.5% and the proportional valve control accuracy within ±0.3%, ensuring a high degree of consistency between the actual flow ratio of each raw material and the optimal proportioning command. Simultaneously, this invention incorporates a model predictive control-driven feedback correction mechanism that continuously monitors the deviation between the actual cooling effect and the predicted target. When the deviation exceeds the allowable range, it automatically triggers incremental adjustments, enabling the system to withstand disturbances such as operating condition fluctuations and parameter drift during long-term operation. This maintains the optimal match between the coolant composition and actual heat dissipation requirements, effectively ensuring the reliability of the battery thermal management system. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of a technical solution for a battery coolant intelligent formulation and component optimization method based on multi-source data fusion. Figure 2 A schematic diagram illustrating the core principle of a multi-source data fusion module that collects multi-dimensional data on battery temperature, current, voltage, and ambient temperature and performs weighted fusion processing. Figure 3 This is a schematic diagram of the three-dimensional parameter performance mapping space structure of the coolant component performance database, covering the multi-dimensional parameter mapping relationship of nanoparticle concentration, base fluid ratio and additive content. Figure 4 This is a schematic diagram illustrating the core principle of a component optimization algorithm that iteratively solves multi-objective optimization problems and outputs the optimal component allocation ratio. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the following description is provided in conjunction with the appendix. Figure 1 To be continued Figure 4 Specific embodiments are provided to further illustrate the present invention in detail.

[0021] This invention provides a method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion. This method achieves real-time dynamic optimization and adjustment of coolant components through the coordinated operation of a multi-source data fusion module, a coolant component performance database, a component optimization algorithm, and a joint control system of a high-precision metering pump and an electric proportional valve, as detailed below.

[0022] The first step, S1, involves multi-source data fusion and thermal load assessment. This step uses a multi-source data fusion module to collect multi-dimensional sensor data, including battery temperature, current, voltage, and ambient temperature. A weighted fusion algorithm is then used for preprocessing and deep fusion to generate a quantitative assessment index that reflects the real-time thermal load status of the battery, providing a data foundation for subsequent dynamic optimization of the coolant composition. This step is implemented through the following sub-steps.

[0023] Step S101: Configure the multi-source data fusion module and establish sensor data acquisition channels. The multi-source data fusion module is configured in the main control unit of the battery thermal management system. The multi-source data fusion module contains four independent data acquisition channels, which establish one-to-one communication connections with the battery temperature sensor, current sensor, voltage sensor, and ambient temperature sensor, respectively.

[0024] Multiple battery temperature sensors are distributed across the surface of each cell in the battery module and in the core area of ​​the module. These sensors utilize digital temperature sensor chips, with a temperature measurement accuracy controlled within ±0.5℃ and a sampling resolution of at least 12 bits. The output signals from each battery temperature sensor are transmitted via a bus interface to the corresponding data acquisition channel of the multi-source data fusion module.

[0025] The current sensor is designed based on the Hall effect principle and is installed at the positive or negative terminal of the DC bus of the battery pack. The current sensor's range covers -300A to +300A, and the sampling frequency is set to 50 times per second. The current sensor's output signal is filtered by a low-pass filter to remove high-frequency ripple interference before being sent to the multi-source data fusion module.

[0026] The voltage sensor is integrated into the main control chip of the battery management system. The voltage sensor sequentially scans the terminal voltage signal of each cell through a multi-channel analog switch, with a sampling frequency set to 20 times per second to ensure the synchronization of voltage data acquisition for each cell.

[0027] The ambient temperature sensor is installed at the air inlet of the battery pack casing. The sensor uses a surface-mount thermistor as the sensing element, with a temperature measurement range of -40℃ to +125℃ and a response time constant of no more than 5 seconds.

[0028] Step S102: The timing alignment module performs time synchronization processing on the raw data of each channel. The raw data acquired by each data acquisition channel must first enter the timing alignment module before entering the fusion calculation. The timing alignment module operates under the global clock of the main control unit, and the global clock period is fixed at 10 milliseconds per cycle.

[0029] The timing alignment module maintains a circular data buffer, the depth of which is set to the least common multiple of the sampling periods of each channel. Since the sampling periods of each channel are different—20 milliseconds for the current channel (50 times per second), 50 milliseconds for the voltage channel (20 times per second), and 100 milliseconds for the temperature channel (synchronized with the fusion output by default)—and the least common multiple of the sampling periods of each channel is 100 milliseconds, the circular data buffer must store at least 10 sets of data to support the historical data required for interpolation calculations.

[0030] During data alignment, data from each channel is resampled according to the nearest timestamp alignment principle. Specifically, the main control unit's master clock is used as a reference, and data collected from different channels at different times are converted to a series of unified time nodes through linear interpolation or nearest-neighbor truncation, thereby eliminating timing misalignment caused by differences in sampling frequency.

[0031] Step S103: The multidimensional data after time-series synchronization is preprocessed and dynamically weighted using the weighted fusion algorithm module. The multidimensional data after time-series alignment is then fed into the weighted fusion algorithm module. The core processing logic of the weighted fusion algorithm module includes three layers: a data preprocessing layer, a dynamic weight calculation layer, and a fusion output layer.

[0032] In the data preprocessing layer, a weighted fusion algorithm performs outlier detection and smoothing on the raw data from each channel. The data preprocessing layer also employs a sliding window filtering algorithm to eliminate random noise. When the variation in data from a certain data acquisition channel exceeds a preset threshold over three consecutive sampling periods, the main control unit determines that the data from that channel exhibits abnormal fluctuations and automatically marks subsequent data acquired from that channel as pending processing.

[0033] In the dynamic weight calculation layer, the weighted fusion algorithm dynamically calculates the current weight coefficients for each data acquisition channel based on the confidence level and historical accuracy of the data from each channel. (Confidence level) The evaluation indicators for (t) include: the effectiveness of the sensor self-test status signal, whether the rate of change of the data acquisition channel data is within a reasonable physical range, and the correlation analysis results of the data acquisition channel data with other data acquisition channel data.

[0034] Confidence The specific determination rules for (t) are as follows: First, regarding the validity of the sensor's self-test status signal, when the self-test status flag returned by the sensor's internal register is normal, the indicator is deemed valid; when the self-test status flag indicates a fault or communication timeout, the indicator is deemed invalid.

[0035] Secondly, regarding the reasonableness of the data change rate, the main control unit calculates the absolute value of the change rate of the current data acquisition channel data over three consecutive sampling periods in real time. When the absolute value of the change rate does not exceed the preset physical change rate threshold corresponding to the physical quantity of the data acquisition channel, the indicator is determined to be normal; when the absolute value of the change rate exceeds the preset physical change rate threshold, the indicator is determined to be abnormal. Specifically, the preset physical change rate threshold is 5 degrees Celsius per second for the temperature channel, 50 amperes per second for the current channel, 10 volts per second for the voltage channel, and 2 degrees Celsius per second for the ambient temperature channel.

[0036] Third, regarding the inter-channel correlation analysis, the main control unit calculates the Pearson correlation coefficient between the current data acquisition channel and the data from other data acquisition channels. When the absolute value of the calculated Pearson correlation coefficient is within the preset reasonable correlation range, the indicator is considered normal; when the absolute value of the Pearson correlation coefficient exceeds the preset reasonable correlation range, the indicator is considered abnormal. Specifically, the reasonable correlation range between the temperature channel and the current channel is 0.50 to 0.90, the reasonable correlation range between the temperature channel and the voltage channel is 0.30 to 0.70, and the reasonable correlation range between the temperature channel and the ambient temperature channel is -0.20 to 0.20.

[0037] Fourth, when all three indicators mentioned above are normal or valid, the confidence level is... (t) takes a value of 1; when any of the above three indicators is abnormal or invalid, the confidence level is... (t) takes the value 0.2.

[0038] The weighting coefficients are updated at the same frequency as the data acquisition frequency; that is, they are recalculated after each data acquisition. Taking a system with n sensor channels as an example, the dynamic weighting coefficient of the k-th data acquisition channel at time t... (t) is determined by the following normalized expression:

[0039] in, (t) represents the real-time confidence level of the k-th data acquisition channel at time t; (t) represents the basic priority coefficient of the Kth data acquisition channel at time t. (t) Based on the contribution of physical quantities in the data acquisition channels to the heat load assessment, the basic priority coefficients are pre-set: temperature channel 0.40, current channel 0.25, voltage channel 0.20, and ambient temperature channel 0.15; denominator This indicates the data acquisition channels. The products are summed, where j is the index variable for traversing each data acquisition channel, and n is the total number of data acquisition channels. In this embodiment, n is 4.

[0040] At the fusion output layer, the acquired data values ​​from each data acquisition channel are... (t) and corresponding weight coefficients Multiplying (t) and summing the results, we obtain the fused battery thermal load assessment index H(t). Wherein, (t) represents the preprocessed data value of the k-th data acquisition channel at time t. The mathematical expression for the battery heat load assessment index H(t) is: H(t) is a dimensionless comprehensive evaluation value. The value range of H(t) is normalized to between 0 and 100. The higher the value, the greater the current thermal load pressure of the battery.

[0041] The weighted fusion algorithm supports adaptive fault tolerance adjustment: when data from a certain data acquisition channel is determined to be abnormal, the confidence level of that data acquisition channel is adjusted. (t) is set to 0.2. After normalization calculation, the final weight of the data acquisition channel will be reduced to less than 20% of the normal value, thereby reducing the interference of abnormal data on the fusion result. At the same time, the main control unit triggers the fault diagnosis mechanism to perform status detection on the sensor of the data acquisition channel.

[0042] In step S104, the main control unit classifies the heat load assessment index into levels according to preset thresholds, and the output data update frequency of the multi-source data fusion module is no less than 10 times per second. The main control unit compares the real-time acquired battery heat load assessment index H(t) with the internally preset multi-level thresholds, and divides the current heat load condition into 5 levels: low load condition, normal load condition, medium load condition, high load condition, and extremely high load condition. Each heat load condition level is pre-associated with and corresponds to a different set of component optimization strategy parameters, providing a clear basis for subsequent coolant component performance database queries and component optimization algorithms to determine the operating condition.

[0043] In step S1, the multi-source data fusion module transforms the raw multi-dimensional sensor data into quantified battery thermal load level information. This battery thermal load level information serves as a key input to the entire intelligent allocation method, directly driving subsequent queries of the coolant component performance database and the execution of component optimization algorithms. This establishes a closed-loop linkage mechanism from operating condition perception to component optimization, providing data prerequisites for real-time dynamic adjustment of coolant components.

[0044] The next step, S2, involves querying the coolant component performance database and extracting candidate proportions. Following the battery heat load level information output in step S1, step S2 uses a pre-built coolant component performance database to filter out candidate proportion ranges that match the current heat load conditions, providing a constrained search space for subsequent component optimization algorithms. This step is implemented through the following sub-steps.

[0045] Step S201: Establish a three-dimensional parameter mapping table structure for the coolant component performance database. The coolant component performance database is stored in the embedded database of the main control unit. The coolant component performance database is organized using a structured three-dimensional parameter mapping table.

[0046] The first dimension of the three-dimensional parameter mapping table is the nanoparticle concentration. The nanoparticle concentration ranges from 0.1% to 2.0%, with a total of 21 concentration gradient nodes, each corresponding to a nanoparticle volume fraction value.

[0047] The second dimension of the three-dimensional parameter mapping table is the base solution ratio. The base solution ratio is defined as the volume percentage combination of water and ethylene glycol. The base solution ratio covers a range from pure water to pure ethylene glycol, with a total of 11 ratio gradient nodes, each corresponding to a water to ethylene glycol volume ratio.

[0048] The third dimension of the three-dimensional parameter mapping table is the additive content. The additive content ranges from 0.01% to 1.0%, with a total of 10 content gradient nodes, each corresponding to a mass fraction value of an additive.

[0049] The gradient nodes in the three dimensions intersect and combine to form a complete three-dimensional parameter space. The coolant component performance database stores performance parameter mapping data for more than 2,300 component ratio schemes.

[0050] Step S202: Configure the performance parameter fields for each record in the coolant component performance database. Each record in the coolant component performance database contains the following key performance parameter fields.

[0051] Thermal conductivity, measured in watts per meter Kelvin, characterizes the heat transfer capacity of a coolant.

[0052] Dynamic viscosity, measured in millipascals per second, characterizes the flow resistance properties of coolant during pumping.

[0053] Cooling efficiency index is a dimensionless parameter that comprehensively characterizes the overall heat dissipation performance of coolant under specific operating conditions.

[0054] Corrosion rate, measured in millimeters per year, characterizes the strength of the corrosive effect of coolant on metallic materials.

[0055] Density, measured in grams per cubic centimeter, is used for mass flow rate conversion.

[0056] Specific heat capacity, measured in joules per gram Kelvin, is used in heat capacity calculations.

[0057] The comprehensive score is a dimensionless parameter. It is a quantitative value of comprehensive performance obtained by weighting and comprehensively evaluating thermal conductivity, dynamic viscosity, cooling efficiency index, corrosion rate, density and specific heat capacity.

[0058] Step S203: A database of coolant component performance is established through systematic experimental testing. This database is created through systematic experimental testing. The experimental testing platform includes a constant-temperature circulating water bath, a precision viscometer, a thermal conductivity meter, a corrosion testing tank, and a data acquisition and analysis system.

[0059] The experimental test was divided into three stages.

[0060] The first stage involved testing the concentration characteristics of the nanoparticles. In this stage, the base solution ratio was kept constant at 50% water and 50% ethylene glycol, and the additive content was fixed at 0.1%. Various performance parameters were measured sequentially at different nanoparticle concentrations to generate a curve showing the relationship between nanoparticle concentration and performance parameters.

[0061] The second stage is the testing of the basic solution ratio characteristics. In the second stage, the concentration of nanoparticles is kept at 0.5% and the additive content is 0.1%. Various performance parameters are measured sequentially under different basic solution ratios to form a curve showing the correspondence between the basic solution ratio and the performance parameters.

[0062] The third stage involves testing the additive content characteristics. In this stage, the nanoparticle concentration was kept at 0.5%, and the base solution ratio was 50% water and 50% ethylene glycol. Various performance parameters were measured sequentially under different additive contents to form a curve showing the relationship between additive content and performance parameters.

[0063] The experimental data from the first, second, and third stages were processed through curve fitting and interpolation to establish a complete three-dimensional parameter performance mapping space. Specifically, cubic spline interpolation was used to fit the discrete measurement data points of the three stages, generating continuous smooth curves between adjacent gradient nodes. Subsequently, the three fitted curves were extended by Cartesian product in the three-dimensional parameter space. For any given combination of nanoparticle concentration, base solution ratio, and additive content, the corresponding predicted performance parameters were obtained from the fitted surface through three-dimensional linear interpolation.

[0064] Through steps S201 to S203, this embodiment completes the construction of the coolant component performance database. The coolant component performance database is organized around a three-dimensional parameter mapping table and based on systematic experimental data, providing complete data support for subsequent operating condition matching queries.

[0065] Step S204: Based on the current heat load condition level, the candidate group allocation ratio range is selected from the three-dimensional parameter performance mapping space. Based on the battery heat load evaluation data obtained after fusion in step S1, the database query module selects the candidate group allocation ratio range that meets the performance requirements from the three-dimensional parameter performance mapping space according to the current heat load condition level information.

[0066] The specific screening strategy is as follows.

[0067] When the heat load assessment index is under low load conditions, the database query module selects the proportion scheme with low thermal conductivity and low dynamic viscosity as candidates, with the main optimization direction being to reduce pumping energy consumption.

[0068] When the heat load assessment index is under high load or extremely high load conditions, the database query module selects ratio schemes with medium to high thermal conductivity and high cooling efficiency index as candidates, with improving heat dissipation capacity as the main optimization direction.

[0069] The extracted candidate group proportion ranges are output in the form of three-dimensional coordinate intervals, which define the upper and lower limits of nanoparticle concentration, basic solution proportion, and additive content, respectively. These three-dimensional coordinate intervals serve as search space constraints for subsequent component optimization algorithms.

[0070] In step S204, the database query module transforms the current heat load level information into a specific, numerical candidate ratio search interval. The output of the search interval indicates that the operating condition perception information has been successfully transformed into constraints for component optimization, providing clear boundary limits for the iterative solution of the component optimization algorithm in step S3.

[0071] In step S3 of the method of this embodiment, the multi-objective solution and optimal proportion output of the component optimization algorithm, taking the three-dimensional coordinate range of the candidate component proportion range output in step S2 as the input constraint, calls the component optimization algorithm deployed in the main control unit processor to perform multi-objective optimization calculation. Under the premise of satisfying the three optimization objectives of improving cooling efficiency, reducing pumping energy consumption, and cost constraints, the optimal component proportion under the current operating conditions is solved. This step is specifically implemented through the following sub-steps.

[0072] Step S301: Configure the input parameter set and output parameter definition of the component optimization algorithm. The component optimization algorithm is deployed in the processor of the main control unit. The input parameters of the component optimization algorithm include the following parameters: the three-dimensional boundary constraints of the candidate component allocation ratio range output in step S204, the current heat load condition level output in step S104, the current ambient temperature value collected by the ambient temperature sensor, the rated power parameters of the cooling system pump, and the unit cost data of each component's raw liquid.

[0073] The component optimization algorithm outputs the optimal component ratio under the current operating conditions. The optimal component ratio specifically includes the optimal values ​​for nanoparticle concentration, base liquid ratio, and additive content.

[0074] Step S302: Construct the multi-objective function system of the component optimization algorithm. The objective function system of the component optimization algorithm includes three optimization objectives.

[0075] The primary optimization objective is to improve cooling efficiency. This objective is quantified by calculating the ratio of the convective heat transfer coefficient of the coolant at the candidate mix ratio to that at the baseline mix ratio. Cooling efficiency improvement rate. The mathematical expression is: in, The convective heat transfer coefficient of the coolant under the candidate group distribution ratio. The unit is watts per square kelvin; This represents the convective heat transfer coefficient of the coolant at the baseline mix ratio. The unit is watts per square kelvin. The first optimization objective requires... ≥1.15, meaning the cooling efficiency is improved by no less than 15%.

[0076] The second optimization objective is to reduce pumping energy consumption. This objective is quantified by calculating the ratio of pump power consumption (based on coolant dynamic viscosity) under the candidate mix ratio to the pump power consumption under the baseline mix ratio. (Pumping energy consumption ratio) The mathematical expression is:

[0077] in, The pump power consumption value calculated based on the dynamic viscosity of the coolant under the representative candidate group allocation ratio. The unit is watt; This represents the pump power consumption value calculated based on the dynamic viscosity of the coolant under the baseline mix ratio. The unit is watts. The second optimization objective requires... ≤0.90, meaning that pumping energy consumption is reduced by no less than 10%.

[0078] The third optimization objective is cost constraint. The quantification method for this third optimization objective is as follows: multiply the unit price of the nanoparticle stock solution by the volumetric quantity of nanoparticles used, multiply the unit price of the base fluid stock solution by the volumetric quantity of base fluid used, and multiply the unit price of the additive stock solution by the volumetric quantity of additives used. Summing these three products yields the total cost per unit volume of coolant. The mathematical expression is:

[0079] in, The unit price of the nanoparticle stock solution. The unit is yuan per liter; This represents the volumetric amount of nanoparticle stock solution used per unit volume of coolant. The unit is liter; This represents the unit price of the base solution. The unit is yuan per liter; This represents the volume of base fluid stock solution used per unit volume of coolant. The unit is liter; This represents the unit price of the additive concentrate. The unit is yuan per liter; This represents the volumetric amount of additive concentrate used per unit volume of coolant. The unit is liter. The third optimization objective requires... The cost will not exceed the preset upper limit of the cost budget, which is preset based on the economic indicators of the battery thermal management system.

[0080] Step S303: Invoke the improved multi-objective particle swarm optimization algorithm or the multi-objective genetic algorithm for iterative solution. The component optimization algorithm supports two algorithm engines for solution: the improved multi-objective particle swarm optimization algorithm and the multi-objective genetic algorithm. The main control unit selects one of the algorithm engines to execute based on the current computing resource utilization and real-time requirements.

[0081] The workflow of the improved multi-objective particle swarm optimization algorithm includes the following steps.

[0082] Initialize the particle swarm: Randomly generate the initial positions and velocities of several particles in the three-dimensional search space. The three dimensions of the three-dimensional search space correspond to the nanoparticle concentration, the base liquid ratio, and the additive content, respectively.

[0083] Calculate the fitness value for each particle: the fitness value is a weighted comprehensive evaluation of the three optimization objective function values. Weighted comprehensive evaluation value The mathematical expression is:

[0084] in, Weighting coefficients representing the target for improving cooling efficiency; The weighting coefficient representing the target of reducing pumping energy consumption; The weighting coefficients representing cost constraints. , , The sum of the three equals 1; This represents the preset upper limit of the cost budget. The unit is yuan per liter; the meanings of the other letters are the same as in step S302.

[0085] Entering the iterative loop: In each iteration, update the individual optimal position and global optimal position of each particle. A non-dominated sorting mechanism is used to perform hierarchical evaluation of the particle swarm, selecting the non-dominated solutions in the first level to form the current optimal solution set. The distribution density of solutions in the optimal solution set is calculated using crowding distance to maintain the diversity of the optimal solution set.

[0086] An adaptive inertial weight is introduced to adjust the particle's flight speed: the inertial weight is reduced when the particle approaches the optimal solution to improve local search accuracy, and increased when the particle moves away from the optimal solution to enhance global exploration capability. The iteration termination condition is reaching the preset maximum number of iterations or the change in the global optimal solution between two consecutive iterations being less than a preset convergence threshold.

[0087] The multi-objective genetic algorithm employs the Non-Dominated Sorting Genetic Algorithm (NSGA-II). Its basic process is as follows: initialize the parent population; perform non-dominated sorting on the individuals; generate the offspring population through crossover and mutation operations; merge the parent and offspring generations and perform non-dominated sorting and crowding distance calculation again; select the best individuals based on the sorting results to form a new generation population; iterate until convergence, and output the Pareto optimal solution set. The specific implementation of this algorithm uses the same evaluation criteria as the improved multi-objective particle swarm optimization algorithm mentioned above regarding the non-dominated sorting and crowding distance maintenance mechanisms, which will not be repeated here.

[0088] Step S304: Select the final optimal group allocation ratio from the Pareto optimal solution set based on the preset decision strategy. After obtaining the Pareto optimal solution set of the component optimization algorithm, the main control unit selects a final implementation scheme from the Pareto optimal solution set as the optimal group allocation ratio for the current working condition according to the preset decision strategy.

[0089] The decision-making strategy employs a weighted summation method, linearly weighting the three objective function values ​​according to preset weight coefficients. The allocation of weight coefficients is dynamically adjusted based on the current heat load level: when the heat load is at a high or extremely high load level, the weight coefficient for the cooling efficiency target is set to the highest value; when the heat load is at a low load level, the weight coefficient for the energy consumption target is set to the highest value.

[0090] In summary, the component optimization algorithm transforms the constraint of the candidate component ratio range output in step S2 into a specific, numerical optimal component ratio. The optimal component ratio includes the optimal values ​​for nanoparticle concentration, base liquid ratio, and additive content. These values ​​serve as execution commands for the high-precision metering pump and electric proportional valve joint control system in subsequent step S4, driving the physical mixing and dynamic adjustment of the coolant components.

[0091] Finally, step S4 involves the dynamic blending and precise mixing control of the coolant components. Following the optimal component ratio output in step S3, a high-precision metering pump and electric proportional valve combined control system physically mixes and dynamically blends the base liquid stock solution, nanoparticle dispersion, and functional additive stock solution stored in the multi-chamber coolant storage system according to the optimal component ratio. This includes the following implementation steps.

[0092] Step S401: Configure the chamber structure and storage medium of the multi-chamber coolant storage system. The multi-chamber coolant storage system includes at least 3 independent storage chambers.

[0093] The first storage chamber is used to store the base fluid stock solution. The base fluid stock solution is a mixture of water and ethylene glycol. The storage capacity of the first storage chamber is determined based on the total capacity of the battery thermal management system's cooling system and the maximum dispensing requirements.

[0094] The second storage chamber is used to store the nanoparticle dispersion. The nanoparticle dispersion is a suspension formed by uniformly dispersing nanoparticles in the base liquid stock solution. The nanoparticles in the nanoparticle dispersion are selected from one or more of alumina, titanium dioxide, or silicon dioxide, and the particle size is controlled within the range of 10 nanometers to 100 nanometers to ensure the long-term stability and uniformity of the nanoparticle dispersion.

[0095] The third reservoir is used to store the functional additive concentrate. The functional additive concentrate contains corrosion inhibitors, scale inhibitors, and corrosion preventants. The functional additive concentrate is used to inhibit the corrosion of metal pipelines by the coolant and prevent scale buildup inside the cooling system.

[0096] Each liquid storage chamber is physically isolated from the others by a sealed partition. The partition is made of corrosion-resistant fluororubber to ensure that different component solutions do not interpenetrate or react chemically during storage.

[0097] Step S402: Configure the liquid level sensing and temperature control mechanism for each liquid storage chamber. Each liquid storage chamber is equipped with an independent liquid level sensor and an independent temperature sensor.

[0098] The level sensor uses a capacitive level gauge and is installed on the side wall of each storage chamber. The level sensor measures the change in liquid level within the storage chamber in real time, with a measurement accuracy controlled within ±2 mm. When the level detected by the level sensor is lower than the preset replenishment threshold, the main control unit automatically triggers a replenishment reminder signal and generates a replenishment command. The replenishment reminder signal is displayed through the human-machine interface, prompting maintenance personnel to replenish the original liquid in a timely manner.

[0099] The temperature sensors are platinum resistance temperature sensors, installed at the bottom of each storage chamber. These sensors monitor the temperature changes of the stored concentrate in real time, with a measurement accuracy controlled within ±0.5℃. When the storage temperature detected by the temperature sensor exceeds the preset safe temperature range, the main control unit automatically activates either the heating or cooling mechanism. The heating mechanism regulates the concentrate temperature through a heating film within the chamber wall, while the cooling mechanism regulates the concentrate temperature through cooling coils within the chamber wall, ensuring the concentrate is stored within the preset optimal storage temperature range.

[0100] Step S403: Configure a high-precision metering pump and execute the basic flow output of each component stock solution. The high-precision metering pump is installed on the outlet pipeline of each storage chamber. The high-precision metering pump is designed using either an electromagnetic drive or a servo motor drive.

[0101] Electromagnetically driven metering pumps control the reciprocating motion of the plunger by switching the power on and off of an electromagnetic coil, achieving precise flow measurement and output. Servo motor driven metering pumps, on the other hand, use a servo motor to drive a ball screw mechanism to move the piston. Servo motor driven metering pumps offer higher positioning accuracy and flow control linearity than electromagnetically driven metering pumps.

[0102] The high-precision metering pump has a response time of no more than 100 milliseconds and completes flow adjustment after receiving control commands from the main control unit. The metering accuracy of the high-precision metering pump is controlled within ±0.5%. The high-precision metering pump uses a built-in closed-loop flow feedback control loop to calibrate the deviation between the actual output flow and the set flow in real time. The closed-loop flow feedback control loop employs a high-speed proportional valve and a precision displacement sensor to form a dual closed-loop control structure.

[0103] A flow sensor is installed on the outlet pipe of the high-precision metering pump. The flow sensor monitors the actual flow rate in real time and feeds it back to the main control unit. The main control unit adjusts the drive signal of the high-precision metering pump based on the flow deviation signal to achieve flow tracking control.

[0104] Step S404: Configure the electric proportional valve and perform fine-tuning and proportional compensation of the flow rates of each component stock solution. The electric proportional valve is installed on the connecting pipeline between the high-precision metering pump and the mixing pipeline. The electric proportional valve adopts a high-speed response design to match the flow adjustment speed of the high-precision metering pump.

[0105] The electric proportional valve employs either a direct-acting or pilot-operated design. Internally, it incorporates a high-speed electromagnet drive mechanism and a precision valve core assembly. The electric proportional valve achieves continuous, stepless adjustment of the valve opening by receiving pulse width modulation (PWM) control signals. The control accuracy of the electric proportional valve is controlled within ±0.3%. The electric proportional valve controls the valve core displacement by varying the duty cycle of the PWM signal, thereby achieving continuous flow regulation.

[0106] The electric proportional valve and the high-precision metering pump work together to form a precise actuator for the component mixing ratio. The high-precision metering pump controls the basic flow output of each component's raw liquid, while the electric proportional valve performs fine adjustments and compensation based on the basic flow. The control signals of the high-precision metering pump and the electric proportional valve are uniformly coordinated and output by the main control unit to ensure that the actual flow ratio of each component remains consistent with the optimal component ratio output in step S3.

[0107] In step S405, the various component stock solutions are uniformly mixed and circulated through the mixing pipeline. The base liquid stock solution, nanoparticle dispersion, and functional additive stock solution, which are metered and output by a high-precision metering pump and finely compensated by an electric proportional valve, converge and are uniformly mixed in the mixing pipeline. The resulting coolant then enters the battery thermal management system for circulation, dissipating heat and cooling the battery module.

[0108] In summary, step S4 transforms the optimal component ratio output in step S3 into a physical mixing operation through a multi-chamber coolant storage system, a high-precision metering pump, and an electric proportional valve combined control system. The base liquid, nanoparticle dispersion, and functional additive liquid are output from independent storage chambers in precise proportions, mixed to form a coolant with the target performance, and then circulated into the battery thermal management system, completing the conversion from a digital optimal ratio command to the actual physical mixing of the coolant.

[0109] After the component ratio adjustment is completed, the intelligent coolant mixing method enters the feedback correction stage.

[0110] The feedback correction mechanism continuously monitors battery temperature trends and thermal management system energy consumption data, with a data acquisition cycle of once per minute. The temperature trend is calculated by using a linear regression method to calculate the rate of temperature change over time from the continuously acquired battery temperature values; this rate of change is then used as the measured value of the battery temperature rise rate. Thermal management system energy consumption data is acquired through a cooling system pump power sensor.

[0111] The feedback correction mechanism compares and analyzes the actual cooling effect with the optimization target, and calculates the deviation value as follows: Cooling efficiency deviation value. The calculation formula is ,in This represents the actual measured rate of battery temperature rise. The model predicts the battery temperature rise rate; energy consumption deviation value. The calculation formula is ,in This represents the actual measured energy consumption value of the thermal management system. This represents the energy consumption value of the thermal management system predicted by the model.

[0112] A deviation signal is generated and fed back to the component optimization algorithm for parameter self-calibration. The specific implementation of parameter self-calibration is as follows: the cooling efficiency deviation value and energy consumption deviation value are used as additional constraint terms and input into the fitness function of the improved multi-objective particle swarm optimization algorithm or multi-objective genetic algorithm in step S303, so that the optimization algorithm automatically adjusts the search direction in the next round of iteration to reduce the deviation value.

[0113] The feedback correction mechanism employs a model predictive control strategy, establishing a cooling effect prediction model based on historical operating data. The cooling effect prediction model is established as follows.

[0114] First, determine the input and output variables of the prediction model. The battery thermal load assessment index generated in step S103, the ambient temperature value collected by the ambient temperature sensor, the coolant flow rate value collected by the coolant flow sensor, and the current component ratio are used as input variables. The battery temperature rise rate and system energy consumption are used as output variables. Let the input variable vector be... ,in This is a battery thermal load assessment index. This is the ambient temperature value. This is the coolant flow rate value. Assign a ratio vector to the current group; let the output variable vector be... ,in For battery temperature rise rate, This refers to the system's energy consumption.

[0115] Second, determine the specific mathematical structure of the multiple linear regression model. The cooling effect prediction model adopts the following multiple linear regression form:

[0116] in, Let be the predicted value of the i-th output variable; This is the regression intercept term corresponding to the i-th output variable; , , , These are the regression coefficients of each input variable corresponding to the i-th output variable; Let be the random error term corresponding to the i-th output variable, which follows a normal distribution with a mean of zero. The above model can be written in matrix form as follows: Where Y is the output variable observation matrix and X is the input variable design matrix. The regression coefficient matrix, This is the error matrix.

[0117] Third, determine the fitting method for the model parameters. (Regression coefficient matrix) The estimation uses ordinary least squares, and the objective function is to minimize the sum of squared residuals:

[0118] Its analytical solution is After fitting, the coefficient of determination is used. The model's goodness of fit was tested, the F-test was used to test the overall significance of the model, and the t-test was used to test the significance of each regression coefficient to ensure the model's effectiveness.

[0119] Fourth, determine the method for collecting and labeling training samples. During the online operation of the battery thermal management system, input variable data for each set are collected at a fixed sampling period, and the measured values ​​of output variables at corresponding times are recorded simultaneously. Each training sample contains a complete set of input variable values ​​and their corresponding measured values ​​of output variables, forming a sample pair (xy). Training sample set It consists of M sets of historical sample pairs, where M is the total number of samples in the current training sample set. Each time a new sample pair is collected, it is added to the training sample set.

[0120] Fifth, determine the online update mechanism for the model. The cooling effect prediction model continuously accumulates new training samples during online operation, triggering a refit of model parameters every 50 new samples. During refitting, all samples in the current training sample set are used as training data, and the regression coefficient matrix is ​​recalculated using the least squares method described above. This allows the model parameters to track the slow changes in the characteristics of the battery thermal management system. After refitting, the updated model parameters replace the original model parameters for subsequent cooling effect prediction.

[0121] When the prediction deviation of the cooling effect prediction model exceeds the allowable range, the main control unit automatically triggers a new round of optimization calculation. The allowable range is set as follows: the allowable range for cooling efficiency deviation is ±5%, and the allowable range for energy consumption deviation is ±5%. When either the cooling efficiency deviation value or the energy consumption deviation value exceeds the corresponding allowable range, a new round of optimization calculation is triggered.

[0122] The feedback correction mechanism supports an incremental adjustment strategy, making only minor adjustments based on the current ratio each time an optimization calculation is triggered. The magnitude of these minor adjustments is limited as follows: the adjustment step size for nanoparticle concentration does not exceed ±5% of the current value; the adjustment step size for base solution ratio does not exceed ±3% of the current value; and the adjustment step size for additive content does not exceed ±5% of the current value. The specific value of the adjustment step size is automatically determined by the component optimization algorithm in step S303 based on the current degree of deviation each time an optimization calculation is triggered.

[0123] During long-term operation, the feedback correction mechanism enables the coolant composition to be adaptively and dynamically adjusted, bringing the system operating state closer to the optimal performance state.

[0124] For ease of understanding, the following is a specific application scenario description of this embodiment. Assume that during the operation of the battery thermal management system, the multi-source data fusion module collects the following data: the average surface temperature sensor value of the battery module is 45 degrees Celsius, the battery discharge current sensor value is 150 amperes, the battery terminal voltage sensor value is 350 volts, and the ambient temperature sensor value is 30 degrees Celsius. The timing alignment module aligns the data from the above four channels according to a unified timestamp and then sends it to the weighted fusion algorithm module. The dynamic weight calculation module performs confidence analysis on the data from each channel according to the confidence level C_k(t) judgment rules disclosed in step S103, determines the weight allocation of each channel according to the normalized expression of the weight coefficient, and after fusion calculation, obtains a battery thermal load assessment index of 75, corresponding to a medium load condition level.

[0125] The database query module filters candidate component ratio ranges from the three-dimensional parameter mapping space based on the heat load level: the candidate range for nanoparticle concentration is 0.5% to 1.0%, the candidate range for base fluid ratio is 40% to 60% water, and the candidate range for additive content is 0.5% to 0.8%. The component optimization algorithm performs multi-objective optimization based on the above candidate ranges, setting three objective functions: improving cooling efficiency, reducing energy consumption, and cost constraints. The optimal component ratio output by the optimization algorithm is: nanoparticle concentration 0.75%, base fluid ratio of 50% water and 50% ethylene glycol, and additive content 0.65%.

[0126] After receiving the optimal mixing ratio command, the high-precision metering pump and electric proportional valve joint control system outputs the raw liquid of each component according to the set flow rate. The electric proportional valve adjusts the valve opening after receiving the pulse width modulation signal to achieve precise mixing. The mixed coolant then circulates in the battery thermal management system. After 15 minutes of operation, the feedback correction mechanism collects and analyzes battery temperature data to confirm that the average surface temperature of the battery module has dropped below 40 degrees Celsius, the cooling efficiency target has been achieved, and the system enters the steady-state operation phase.

[0127] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0128] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion, characterized in that, include: The multi-source data fusion module collects multi-dimensional sensor data of battery temperature, current, voltage and ambient temperature. The weighted fusion algorithm is used to preprocess and deeply fuse the multi-dimensional sensor data to generate a battery thermal load assessment index that reflects the real-time thermal load status of the battery. The current thermal load condition is divided into different levels according to a preset threshold. Based on the level of the current heat load condition, candidate component ratio ranges that match the current heat load condition are selected from a pre-built coolant component performance database; wherein, the coolant component performance database is a three-dimensional parameter mapping table covering three dimensions: nanoparticle concentration, base fluid ratio, and additive content. Using the candidate group allocation ratio range as a search space constraint, the component optimization algorithm is invoked to perform multi-objective optimization calculation. The optimization objectives of the multi-objective optimization calculation include at least improving cooling efficiency, reducing pumping energy consumption, and cost constraints. The optimal group allocation ratio under the current operating conditions is solved and output. The system uses a high-precision metering pump and an electric proportional valve to physically mix and dynamically adjust the base liquid stock solution, nanoparticle dispersion and functional additive stock solution stored in the multi-chamber coolant storage system according to the optimal composition ratio.

2. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 1, characterized in that, The generation of the battery thermal load assessment index, which reflects the real-time thermal load status of the battery, specifically includes: The timing alignment module resamples the raw data collected by each data acquisition channel under the global clock drive within the timing alignment module, according to the nearest timestamp alignment principle, in order to eliminate timing misalignment caused by sampling frequency differences. The multidimensional data after time-series synchronization is processed by a weighted fusion algorithm module, which includes a data preprocessing layer, a dynamic weight calculation layer, and a fusion output layer. In the dynamic weight calculation layer, the current weight coefficient for each data acquisition channel is dynamically calculated based on the confidence level and historical accuracy of the data from each data acquisition channel. The confidence level evaluation indicators include the validity of the sensor self-test status signal, whether the rate of change of the data acquisition channel data is within a reasonable physical range, and the correlation analysis results between the data acquisition channel data and other data acquisition channels. When the data from a certain data acquisition channel is determined to be abnormal, the confidence level of the abnormal data acquisition channel is set to 0.2, causing the final weight of the abnormal data acquisition channel after normalization calculation to drop below 20% of the normal value. In the fusion output layer, the collected data values ​​of each data acquisition channel are multiplied by their corresponding weight coefficients and then summed to generate the fused battery thermal load evaluation index.

3. The intelligent formulation and component optimization method for battery coolant based on multi-source data fusion according to claim 1, characterized in that, The steps for establishing the coolant component performance database include: A three-dimensional parameter mapping table is established. The first dimension of the three-dimensional parameter mapping table is the nanoparticle concentration, covering multiple concentration gradient nodes from 0.1% to 2.0%. The second dimension of the three-dimensional parameter mapping table is the base liquid ratio, which is defined as the volume percentage combination of water and ethylene glycol, covering multiple ratio gradient nodes from pure water to pure ethylene glycol. The third dimension of the three-dimensional parameter mapping table is the additive content, covering multiple content gradient nodes from 0.01% to 1.0%. The gradient nodes of the three dimensions are cross-combined to form a three-dimensional parameter space. For each group allocation scheme in the three-dimensional parameter space, a corresponding performance parameter field is configured. The performance parameter field includes at least thermal conductivity, dynamic viscosity, cooling efficiency index, and comprehensive score.

4. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 1, characterized in that, The invocation of the component optimization algorithm to perform multi-objective optimization calculations and solve for the optimal component allocation ratio under the current operating conditions specifically includes: The component optimization algorithm is solved iteratively using an improved multi-objective particle swarm optimization algorithm or a multi-objective genetic algorithm; The improved multi-objective particle swarm optimization algorithm introduces an adaptive inertial weight to adjust the flight speed of particles during the iteration process. The adaptive inertial weight is reduced when the particles approach the optimal solution and increased when the particles move away from the optimal solution. After obtaining the Pareto optimal solution set, the final implementation scheme is selected from the Pareto optimal solution set as the optimal group allocation ratio according to the preset decision strategy; wherein, the decision strategy adopts the weighted summation method, and the weight coefficient of the weighted summation method is dynamically adjusted according to the level of the current heat load condition. When the heat load is at the high load condition level or the extremely high load condition level, the weight coefficient of the cooling efficiency target is set to the highest value.

5. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 1, characterized in that, The physical mixing and dynamic adjustment achieved through a combined control system of a high-precision metering pump and an electric proportional valve specifically includes: The high-precision metering pump is installed on the outlet pipe of each storage chamber of the multi-chamber coolant storage system. The response time of the high-precision metering pump is no more than 100 milliseconds, and the metering accuracy of the high-precision metering pump is controlled within ±0.5%. The high-precision metering pump calibrates the deviation between the actual output flow and the set flow in real time through a built-in closed-loop flow feedback control loop. The electric proportional valve is installed on the connecting pipeline between the high-precision metering pump and the mixing pipeline. The control accuracy of the electric proportional valve is controlled within ±0.3%. The electric proportional valve realizes continuous stepless adjustment of the valve opening by receiving pulse width modulation control signal. The high-precision metering pump controls the basic flow output of the base liquid stock solution, nanoparticle dispersion, and functional additive stock solution. The electric proportional valve performs flow fine-tuning and compensation based on the basic flow controlled by the high-precision metering pump. The main control unit coordinates and outputs the control signals of the high-precision metering pump and the electric proportional valve to ensure that the actual flow ratio of the base liquid stock solution, nanoparticle dispersion, and functional additive stock solution remains consistent with the optimal composition ratio.

6. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 1, characterized in that, The multi-chamber coolant storage system comprises at least three independent storage chambers. The first storage chamber stores a base liquid stock solution composed of a mixture of water and ethylene glycol. The second storage chamber stores a nanoparticle dispersion formed by uniformly dispersing nanoparticles in the base liquid stock solution. The third storage chamber stores a functional additive stock solution containing corrosion inhibitor, scale inhibitor, and preservative components. Each storage chamber is equipped with an independent capacitive level gauge and an independent platinum resistance temperature sensor to monitor the changes in liquid level and temperature of the stored stock solution in real time.

7. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 1, characterized in that, After adjusting the group allocation ratio, a feedback correction mechanism is also included, which specifically includes: Continuously monitor battery temperature change trends and thermal management system energy consumption data, and calculate the cooling efficiency deviation and energy consumption deviation between the actual cooling effect and the optimization target; A model predictive control strategy is adopted, and a cooling effect prediction model is established based on historical operating data. The battery thermal load assessment index, the ambient temperature value collected by the ambient temperature sensor, the coolant flow rate value collected by the coolant flow sensor, and the current component ratio are used as input variables, and the battery temperature rise rate and system energy consumption are used as output variables. The cooling effect prediction model adopts a multiple linear regression model. When either the cooling efficiency deviation or the energy consumption deviation exceeds the preset allowable range, the main control unit automatically triggers a new round of optimization calculation.

8. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 7, characterized in that, The feedback correction mechanism supports an incremental adjustment strategy. Each time a new round of optimization calculation is triggered, only a small adjustment is made based on the current ratio. Specifically, the adjustment step size for nanoparticle concentration does not exceed ±5% of the current nanoparticle concentration value, the adjustment step size for base liquid ratio does not exceed ±3% of the current base liquid ratio value, and the adjustment step size for additive content does not exceed ±5% of the current additive content value. The specific values ​​of the adjustment step size for nanoparticle concentration, base liquid ratio, and additive content are automatically determined by the component optimization algorithm based on the current cooling efficiency deviation value and energy consumption deviation value.

9. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 3, characterized in that, The establishment of the coolant component performance database was completed through systematic experimental testing, specifically including: In the first phase of testing, the base solution ratio and additive content were kept constant, and various performance parameters were measured sequentially at different nanoparticle concentrations to form a curve showing the relationship between nanoparticle concentration and performance parameters. In the second phase of testing, while keeping the concentration of nanoparticles and the content of additives constant, various performance parameters were measured sequentially under different base solution ratios to form a curve showing the relationship between the base solution ratio and the performance parameters. In the third stage of testing, while keeping the concentration of nanoparticles and the ratio of base solution constant, various performance parameters were measured sequentially under different additive contents to form a curve showing the relationship between additive content and performance parameters. For the discrete measurement data points obtained from the first stage test, the second stage test, and the third stage test, the cubic spline interpolation method is used to perform curve fitting, and the three fitted curves are extended by Cartesian product in the three-dimensional parameter space to form a complete three-dimensional parameter performance mapping space.

10. The method for intelligent formulation and component optimization of battery coolant based on multi-source data fusion according to claim 3, characterized in that, The process of selecting candidate component ratio ranges that match the current heat load conditions from a pre-built coolant component performance database specifically includes: When the battery thermal load evaluation index is under low load conditions, the formulation scheme with a low thermal conductivity and low dynamic viscosity is selected as a candidate. When the battery thermal load evaluation index is under high load or extremely high load conditions, the ratio scheme with a medium to high thermal conductivity and a high cooling efficiency index is selected as a candidate.

Images