Battery active pre-cooling path planning method and thermal management system

Abstract

The application discloses a battery active precooling path planning method and a thermal management system, and the method comprises the following steps: S1, obtaining journey-related information of a future journey of a vehicle, performing feature extraction on the journey-related information to obtain a journey feature vector; and simultaneously obtaining current state information of the battery; S2, inputting the journey feature vector and the current state information into a pre-constructed probability prediction model, and outputting, by the probability prediction model, a probability distribution sequence of a battery load power and a probability distribution sequence of an external environment temperature in a future prediction time domain as an uncertainty prediction distribution sequence. Compared with the prior art, the application has the beneficial effects that: the application can solve the problem that the existing deterministic active precooling planning method is difficult to effectively cope with the uncertainty of future working conditions, reduce unnecessary precooling energy consumption while ensuring the thermal safety of the battery, and improve the robustness, economy and real vehicle adaptability of the battery active precooling control.

Description

Technical Field

[0001] This invention belongs to the field of battery thermal management, specifically relating to a battery active pre-cooling path planning method and thermal management system. Background Technology

[0002] Electric vehicle batteries are highly sensitive to operating temperature. In high-temperature environments or under high-power charging and discharging conditions, excessively high battery temperatures can lead to limited output power, reduced charging speed, accelerated battery aging, and even the risk of thermal runaway. Therefore, active pre-cooling of the battery is of great importance.

[0003] Existing active pre-cooling control methods for batteries mostly adopt a deterministic paradigm of prediction followed by planning. That is, based on the navigation path, traffic conditions, ambient temperature and current battery status, the future changes in battery temperature are predicted, and the timing and power of pre-cooling are determined accordingly.

[0004] However, in actual driving, factors such as vehicle speed, road congestion, driving behavior, and weather changes can all introduce uncertainty into future load power and external ambient temperature. If conservative operating conditions are used for prediction, precooling may occur too early or excessively, increasing energy consumption; if ideal operating conditions are used, precooling may be insufficient, making it difficult to balance battery thermal safety and energy efficiency. This means that existing deterministic precooling planning paradigms often struggle to balance battery thermal safety and precooling energy consumption control when dealing with uncertainties in future operating conditions. Therefore, a method is urgently needed that can effectively characterize uncertainty and perform path planning accordingly. Summary of the Invention

[0005] The purpose of this invention is to provide a battery active precooling path planning method and thermal management system to solve the technical problems mentioned in the background above.

[0006] To achieve the above objectives, this invention provides a battery active pre-cooling path planning method, comprising the following steps: S1, acquiring trip-related information of the vehicle's future journey, extracting features from the trip-related information to obtain a trip feature vector; simultaneously acquiring the current state information of the battery; S2, inputting the trip feature vector and the current state information into a pre-constructed probability prediction model, and having the probability prediction model output a probability distribution sequence of battery load power and a probability distribution sequence of external ambient temperature in the future prediction time domain, as an uncertainty prediction distribution sequence; S3, using the uncertainty prediction distribution sequence as input, combined with a preset battery... A thermal state model is used to construct a chance-constrained programming problem. The chance-constrained programming problem uses the pre-cooling start time and the pre-cooling power curve as decision variables, minimizes the expected value of pre-cooling energy consumption as the optimization objective, and is constrained by the probability that the battery temperature is within a preset safe range for all time steps in the prediction time domain under the random scenario represented by the uncertainty prediction distribution sequence is not less than a preset confidence level. S4. The chance-constrained programming problem is solved to obtain a pre-cooling path planning scheme that includes the pre-cooling start time and the pre-cooling power target curve. S5. According to the pre-cooling path planning scheme, the battery pre-cooling system is controlled to perform active pre-cooling on the battery.

[0007] In one or more embodiments of the present invention, in S1, the trip-related information includes navigation path information, traffic situation information, and ambient temperature information; the current status information includes the current battery temperature and the current state of charge.

[0008] In one or more embodiments of the present invention, in S2, the probabilistic prediction model is a probabilistic prediction model based on a sequence-to-sequence architecture, and the probabilistic prediction model includes an encoding module and a probabilistic decoding module; the step of inputting the trip feature vector and the current state information into the pre-constructed probabilistic prediction model, and having the probabilistic prediction model output the probability distribution sequence of battery load power and the probability distribution sequence of external ambient temperature in the future prediction time domain, includes: the encoding module is used to encode the features of each road segment in the trip feature vector and the current state information to obtain time-series context features; the probabilistic decoding module, based on the time-series context features, outputs the distribution parameters of battery load power and the distribution parameters of external ambient temperature at each time step in the prediction time domain, so as to constitute the probability distribution sequence of battery load power and the probability distribution sequence of external ambient temperature, respectively.

[0009] In one or more embodiments of the present invention, in S3, the battery thermal state model is a battery thermal balance recursive model, which is used to calculate the corresponding battery temperature change based on the battery load power, external ambient temperature and pre-cooling power at each time step, so as to recursively obtain the battery temperature at each time step.

[0010] In one or more embodiments of the present invention, S3 specifically includes: Step S31: parameterizing the pre-cooling start-up time and the pre-cooling power curve into a parameter vector; Step S32: generating a set of scenarios representing multiple possible future operating conditions through Monte Carlo sampling based on the probability distribution sequence; Step S33: for each scenario in the scenario set, using the battery thermal balance recursive model, calculating the battery temperature at each time step in the prediction time domain under that scenario based on the battery load power time series, the external ambient temperature time series, and the pre-cooling start-up time and pre-cooling power curve represented by the parameter vector; Step S34: constructing the expected value of the pre-cooling energy consumption as the mean of the pre-cooling energy consumption under the scenario set; constructing the probability that the battery temperature is within the preset safety range as the proportion of the number of scenarios where the battery temperature is within the preset safety range at all time steps in the prediction time domain to the total number of scenarios in the scenario set; Step S35: constructing the opportunity-constrained programming problem based on the expected value of the pre-cooling energy consumption and the probability that the battery temperature is within the preset safety range.

[0011] In one or more embodiments of the present invention, in S4, solving the chance-constrained planning problem to obtain a precooling path planning scheme including the precooling start time and the precooling power target curve includes: using an iterative optimization algorithm based on random scenario sampling to solve the chance-constrained planning problem; in each iteration, re-executing the Monte Carlo sampling to update the scenario set, and recalculating the expected value of the precooling energy consumption and the probability that the battery temperature is within a preset safe range based on the updated scenario set to update the parameter vector, until the constraint conditions are met and the expected value of the precooling energy consumption converges to obtain the optimal parameter vector; restoring the optimal parameter vector to the precooling start time and the precooling power target curve.

[0012] In one or more embodiments of the present invention, the battery thermal balance recursive model is characterized by the following equation:

[0013]

[0014] in, Let be the battery temperature at time step t. Let be the external ambient temperature at time step t. Let be the battery load power at time step t. Let be the precooling power at time step t. For time step, The equivalent heat capacity of the battery. Let the heat generated by the battery load power be the function. This is used to characterize the heat dissipation power function of the precooling system at the current battery temperature and ambient temperature.

[0015] In one or more embodiments of the present invention, S6 is further included: during the actual driving process of the vehicle, the actual load power and the actual ambient temperature are collected in real time; when the actual load power or the actual ambient temperature falls outside the preset confidence interval of the corresponding time step determined according to the corresponding probability distribution sequence, the travel feature vector and the current state information are updated using the currently collected actual driving data, and steps S2 to S4 are retried to generate and execute the updated pre-cooling path planning scheme.

[0016] In one or more embodiments of the present invention, the preset confidence level increases as the thermal risk level of the future journey increases; the thermal risk level is determined based on at least one of fast charging demand, long uphill sections, and high-temperature congested sections.

[0017] In addition, to implement the above method, this invention provides a thermal management system for implementing an active pre-cooling path planning method for batteries, including an information acquisition module, a feature processing module, a prediction planning module, and a pre-cooling execution control module. The information acquisition module is used to acquire trip-related information of the vehicle's future journey and the current state information of the battery. The trip-related information includes navigation path information, traffic situation information, and ambient temperature information, and the current state information of the battery includes the current battery temperature and the current state of charge. The feature processing module is connected to the information acquisition module and is used to process the trip-related information to obtain a trip feature vector. The prediction planning module and... The feature processing module is connected to obtain the uncertainty prediction distribution sequence of battery load power and external ambient temperature in the future prediction time domain based on the travel feature vector and the current state information of the battery. Based on the uncertainty prediction distribution sequence and the battery thermal state model, it constructs a chance-constrained planning problem to obtain a pre-cooling path planning scheme. The pre-cooling path planning scheme includes the pre-cooling start time and the pre-cooling power target curve. The pre-cooling execution control module is connected to the prediction planning module and generates a pre-cooling control command based on the pre-cooling path planning scheme. Based on the pre-cooling control command, it controls the battery pre-cooling system to perform active pre-cooling on the power battery.

[0018] Compared with the prior art, the beneficial effects of the present invention are: it can effectively characterize the uncertainty of future operating conditions in the existing deterministic active precooling planning method, effectively characterize the uncertainty of future battery load power and external ambient temperature, and generate a precooling path planning scheme based on chance constraint planning, thereby reducing unnecessary precooling energy consumption while ensuring battery thermal safety, and improving the robustness, economy and real vehicle adaptability of active battery precooling control. Attached Figure Description

[0019] Figure 1 This is a flowchart of a battery active precooling path planning method in one embodiment of the present invention;

[0020] Figure 2 This is a flowchart illustrating the acquisition of trip-related information and current battery status information in one embodiment of the present invention.

[0021] Figure 3 This is a flowchart illustrating the formation of a travel feature vector in one embodiment of the present invention;

[0022] Figure 4 This is a flowchart of the probabilistic prediction model in one embodiment of the present invention;

[0023] Figure 5 This is a flowchart illustrating the construction of a chance-constrained programming problem in one embodiment of the present invention.

[0024] Figure 6 This is a schematic diagram illustrating the objective and constraints of a chance-constrained programming problem in one embodiment of the present invention;

[0025] Figure 7 This is a flowchart illustrating the iterative optimization solution of a chance-constrained planning problem in one embodiment of the present invention.

[0026] Figure 8 This is a flowchart illustrating the closed-loop update process during actual driving in one embodiment of the present invention.

[0027] Figure 9 This is a flowchart illustrating the pre-set confidence level for adjusting thermal risk levels in one embodiment of the present invention;

[0028] Figure 10 This is a structural block diagram of a thermal management system in one embodiment of the present invention. Detailed Implementation

[0029] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings, but it should be understood that the scope of protection of the present invention is not limited to the specific embodiments.

[0030] like Figure 1 As shown, a battery active precooling path planning method can be executed by a vehicle controller, a battery thermal management controller, a vehicle controller, or a vehicle domain controller. It is used to proactively plan the battery precooling start time and precooling power target curve based on future trip information, current battery status, and uncertainties in future operating conditions before the vehicle departs or during actual driving, thereby reducing unnecessary precooling energy consumption while ensuring battery thermal safety.

[0031] In practice, the vehicle's battery management system typically collects data on the current battery temperature, current state of charge, battery current, battery voltage, and multiple temperature measurement points within the battery pack. The navigation module provides information such as the future route, road gradient, road type, speed limit, and estimated arrival time. The communication module receives external information such as traffic conditions, weather forecasts, ambient temperature, and charging station status. The actuator implementing this method can adjust the compressor speed, electric water pump speed, valve opening, or target coolant temperature based on the pre-cooling path planning scheme obtained by this method to achieve active pre-cooling of the power battery.

[0032] It is worth noting that the pre-cooling planning in this invention is not continuously initiated at any given time, but rather executed when pre-cooling decision trigger conditions are met. Trigger conditions may include: the user setting a navigation destination; the vehicle about to enter a high-temperature, congested road section; the future route including a long uphill section; anticipated arrival at a fast-charging station with fast-charging demand; the current battery temperature approaching a preset thermal risk threshold; or a judgment based on the travel characteristics over a future period indicating a high demand for charging and discharging power. By setting these trigger conditions, meaningless execution of complex predictions and planning in low-thermal-risk scenarios can be avoided, thereby reducing onboard computing overhead and pre-cooling energy consumption.

[0033] like Figure 2 As shown, in step S1, the trip-related information of the vehicle's future journey is obtained, and the trip-related information is used to extract features to obtain a trip feature vector; at the same time, the current state information of the battery is obtained.

[0034] The trip-related information includes navigation route information, traffic situation information, and ambient temperature information. Navigation route information can be generated by the in-vehicle navigation module or distributed by a cloud-based navigation service via the vehicle's communication unit. This information may include the origin, destination, waypoints, route mileage, road gradient, speed limit, road type, and estimated arrival time. Traffic situation information may include real-time traffic speed, congestion level, road construction information, traffic light density, or estimated transit time for each road segment. Ambient temperature information can come from a meteorological service interface or be obtained by fusing weather forecast data along the route with values ​​collected by the in-vehicle ambient temperature sensor. Battery current status information includes the current battery temperature and current state of charge; it may also include the current battery current, battery voltage, highest temperature measurement point, lowest temperature measurement point, and battery pack temperature difference.

[0035] like Figure 3As shown, during feature extraction, the future journey is divided into multiple sequentially arranged prediction units based on the navigation path. Prediction units can be divided according to fixed time steps, such as every 10 to 120 seconds; or according to fixed road segment lengths, such as every 100 to 1000 meters; or adaptively based on road type changes, significant gradient changes, or speed limit changes. For each prediction unit, features such as road speed limit, road segment length, gradient, congestion index, estimated travel time, and ambient temperature are extracted and organized into a journey feature vector according to the vehicle's travel sequence.

[0036]

[0037] in, This represents the road segment characteristics of the i-th prediction unit. Indicates the number of prediction units. This indicates the current status information. The current status information includes at least the current battery temperature and the current state of charge, and may further include the current battery current, battery voltage, maximum battery pack temperature, minimum battery pack temperature, or module temperature difference.

[0038] Preferably, before forming the trip feature vector, data from different sources are time-stamp aligned and confidence-filtered. For example, traffic situation information is matched with navigation paths according to road segment numbers, hourly weather forecast data is interpolated to the corresponding prediction units based on the vehicle's expected arrival time at each road segment, and battery status information is low-pass filtered or filtered through a sliding window based on the sampling time. For outliers that clearly exceed physical boundaries, such as instantaneous changes in ambient temperature exceeding a preset rate of change limit, the mean of nearby time steps or the most recent reliable value can be used as a substitute. This processing can improve the consistency of input data and reduce the impact of noise on subsequent probability prediction results.

[0039] Step S1 transforms information from different dimensions—originally scattered across navigation, traffic, weather, and battery management systems—into a unified input that can be directly processed by subsequent probabilistic prediction models. Since future battery load power and ambient temperature are not determined by a single factor but are jointly related to road gradient, traffic congestion, projected speed, external temperature, and current battery status, this feature extraction process provides a more complete data foundation for subsequent processing steps.

[0040] like Figure 4 As shown, in step S2, the travel feature vector and the current state information are input into a pre-constructed probability prediction model. The probability prediction model outputs the probability distribution sequence of the battery load power and the probability distribution sequence of the external environment temperature in the future prediction time domain, which serve as the uncertainty prediction distribution sequence.

[0041] It is worth noting that this scheme does not only predict a single load power value or a single ambient temperature value at a future time step, but also outputs the possible range of values ​​for that time step and their probability characteristics. This can explicitly characterize the uncertainties brought about by changes in road congestion, differences in driving behavior, traffic light waiting, road conditions on slopes, and weather changes, providing a basis for risk decision-making in subsequent opportunity-constrained planning.

[0042] Preferably, the probabilistic prediction model adopts a sequence-to-sequence architecture-based probabilistic prediction model, including an encoding module and a probabilistic decoding module. The encoding module jointly encodes the features of each prediction unit in the travel feature vector with the current state information to obtain temporal context features; the probabilistic decoding module, based on the temporal context features, outputs the battery load power distribution parameters and external ambient temperature distribution parameters for each time step within the prediction time domain. Its output can be expressed as:

[0043]

[0044] in, This represents the temporal context feature corresponding to the t-th time step. This represents the probability decoding mapping function. and Let represent the mean and standard deviation of the battery load power distribution at time step t, respectively. and Let represent the mean and standard deviation of the external ambient temperature distribution at time step t, respectively.

[0045] Specifically, battery load power and external ambient temperature can be modeled as Gaussian distributions:

[0046]

[0047]

[0048] in, This represents the battery load power at time step t. This represents the ambient temperature at time step t. The mean characterizes the central tendency of the corresponding operating condition variable, while the standard deviation characterizes the magnitude of the prediction uncertainty. A larger standard deviation indicates a wider range of future operating condition fluctuations corresponding to that time step, and a lower level of prediction confidence.

[0049] Preferably, when the future operating conditions of a vehicle exhibit significant asymmetric or multimodal characteristics, the probabilistic prediction model can also employ quantile regression models, Gaussian mixture models, Bayesian neural network models, or ensemble learning models to output probability distribution sequences. For example, on the same road segment, a vehicle may exhibit different modes such as smooth cruising, stop-and-go traffic, and short-term rapid acceleration. Mixed distribution models can more fully represent these multimodal operating conditions than a single Gaussian distribution.

[0050] The probabilistic prediction model can be trained offline using historical operational data before vehicle deployment. Training data may include historical navigation routes, historical traffic conditions, historical ambient temperatures, battery status, actual vehicle speed, actual battery load power, and actual ambient temperature. After training, the model parameters are stored in the onboard controller or cloud-based prediction module, allowing for real-time inference based on current trip characteristics during vehicle operation.

[0051] like Figure 5 As shown, in step S3, the uncertainty prediction distribution sequence is used as input, and a chance-constrained programming problem is constructed in conjunction with a preset battery thermal state model. The chance-constrained programming problem uses the pre-cooling start-up time and the pre-cooling power curve as decision variables, minimizes the expected value of pre-cooling energy consumption as the optimization objective, and is constrained by the probability that the battery temperature is within a preset safe range for all time steps in the prediction time domain under the random scenario represented by the uncertainty prediction distribution sequence is not less than a preset confidence level.

[0052] It should be noted that the constraints in this step require the battery temperature to remain within a preset safe range for all time steps within the prediction time domain. This is a joint chance constraint for the entire temperature trajectory. Compared to the point-by-point approach that only requires each individual time step to meet the safety constraints, the joint chance constraint focuses on whether any time step in the entire prediction time domain will exceed the limit, thus providing more reliable control over the overall thermal risk.

[0053] Specifically, step S3 may include steps S31 to S35.

[0054] In step S31, the precooling start-up time and the precooling power curve are parameterized together into a parameter vector:

[0055]

[0056] in, Represents the decision parameter vector. Indicates the pre-cooling start-up time. to This represents the power value of the control node or control segment of the precooling power curve. This indicates the number of segments in the precooling power control. The precooling power curve can be generated using piecewise constant, piecewise linear, or spline interpolation methods. For time steps before the precooling start-up time, the precooling power can be zero or maintain the vehicle's basic thermal management power; for time steps at and after the precooling start-up time, the precooling power is obtained by expanding the control nodes in the parameter vector.

[0057] By incorporating the precooling start-up time and precooling power curve into a single parameter vector, it is possible to determine "when to start precooling" and "at what power" within the same optimization framework, making it easier for the precooling strategy to achieve a global balance between thermal safety and energy economy.

[0058] Preferably, to accommodate the physical limitations of the vehicle's thermal management actuators, the precooling power can also be constrained by maximum power, minimum power, and rate of change. For example, the compressor speed, electric water pump speed, and electronic expansion valve opening are all limited to their respective allowable ranges, and the adjustment range between adjacent time steps must not exceed the allowable rate of change of the actuator. This setting can prevent frequent and significant speed adjustments of the compressor, reduce noise, vibration, and actuator wear, and improve the feasibility of the precooling path planning scheme.

[0059] In step S32, based on the probability distribution sequence output in step S2, a set of scenarios representing various possible future working conditions is generated through Monte Carlo sampling. For the first... For a random scenario, samples are taken from the probability distributions of battery load power and external ambient temperature at each time step within the prediction time domain to obtain a battery load power time series and an external ambient temperature time series. This random scenario can be represented as:

[0060]

[0061]

[0062] in, Let j represent the j-th random scenario. This represents the battery load power time series in the prediction time domain under this scenario. This represents the time series of external environmental temperature in the prediction time domain under this scenario. This indicates the number of time steps within the prediction time domain. Indicates the number of random scenes.

[0063] Number of scenes The number of scenarios can be set according to the computing power of the vehicle controller and the real-time requirements of the planning, for example, it can be set from 100 to 1000; on platforms with strong computing power, it can also be increased to thousands of scenarios to enhance the stability of probability estimation.

[0064] Preferably, on resource-constrained vehicle platforms, Latin hypercube sampling, stratified sampling, or low-discrepancy sequence sampling can be employed to reduce the number of samples required to achieve the same probability estimation accuracy. By transforming continuous probability distributions into discrete random scenarios, uncertain operating conditions can be directly calculated using the battery thermal state model, thereby enhancing the engineering feasibility of the chance-constrained programming problem.

[0065] In step S33, for each scenario in the scenario set, the battery temperature at each time step in the prediction time domain is recursively calculated using the battery thermal balance recursive model, based on the battery load power time series, the external ambient temperature time series, and the pre-cooling start-up time and pre-cooling power curve represented by the parameter vector in that scenario.

[0066] In this embodiment, the battery thermal state model adopts a battery thermal balance recursive model. To maintain consistency with the model form in the claims, the battery thermal balance recursive model itself does not change due to different random scenarios; the only difference between different scenarios lies in the battery load power sequence and the external ambient temperature sequence substituted into the model. The battery thermal balance recursive model can be expressed as:

[0067]

[0068] in, For the first Battery temperature at each time step. Let be the battery temperature at time step t. Let be the battery load power at time step t. Let be the external ambient temperature at time step t. Let be the precooling power at time step t. For time step, The equivalent heat capacity of the battery. Let the heat generated by the battery load power be the function. This is used to characterize the heat dissipation power function of the precooling system at the current battery temperature and ambient temperature.

[0069] In the In a random scenario, the sampled battery load power time series and external ambient temperature time series are used as the parameters in the above model. and At the same time, the parameter vector The precooling start-up time and precooling power curves represented are reconstructed for each time step. This is then substituted into the aforementioned battery thermal balance recursive model. Thus, starting from the current battery temperature, the battery temperature trajectory in the predicted time domain under this random scenario is obtained by recursively extrapolating according to the time steps.

[0070] The physical meaning of this model is that the battery temperature in the next time step equals the current temperature plus the temperature change caused by the difference between heat generation and heat dissipation. When the battery load power is high, the output of the heat generation function increases, and the battery temperature rise trend intensifies; when the pre-cooling power increases, the output of the heat dissipation power function increases, and the battery temperature rise is suppressed; when the external ambient temperature rises, the heat exchange capacity of the pre-cooling system may decrease, increasing the difficulty of battery temperature control. Therefore, this model can uniformly map load power, ambient temperature, and pre-cooling power into battery temperature changes, providing a calculation basis for subsequent judgments on whether the battery temperature is within the preset safe range.

[0071] Preferably, the battery heat generation function can be obtained through the battery equivalent internal resistance model, load power fitting model, calibration lookup table model or data-driven regression model; the pre-cooling heat dissipation power function can be obtained through the calibration relationship between pre-cooling power, coolant inlet temperature, battery temperature, ambient temperature and heat transfer coefficient.

[0072] Preferably, a liquid cooling plate can be installed at the bottom of the battery pack. The liquid cooling plate is made of aluminum alloy and has serpentine flow channels, parallel flow channels, or microchannel flow channels inside. The height of the flow channels can be 1 mm to 5 mm. The liquid cooling plate contacts the bottom of the battery module through thermally conductive pads, which can be made of thermally conductive silicone sheets, thermally conductive gels, or phase change thermally conductive materials. The coolant can be an aqueous solution of ethylene glycol to balance low-temperature antifreeze performance and heat exchange performance. The above structural configuration can improve the heat exchange efficiency of the pre-cooling system for the battery pack, giving the planned pre-cooling power target curve a clear execution path.

[0073] In step S34, the average pre-cooling energy consumption under the scenario set is used to construct the expected value of pre-cooling energy consumption; at the same time, the proportion of the number of scenarios in which the battery temperature is within the preset safe range at all time steps in the prediction time domain to the total number of scenarios in the scenario set is used to construct the probability that the battery temperature is within the preset safe range.

[0074] Let the preset safe zone be ,in, This is the lower limit of the normal operating temperature of the battery. This refers to the upper limit of the battery's safe operating temperature. For active pre-cooling scenarios, the main concern is whether the battery temperature exceeds this limit. But retain This can avoid excessive pre-cooling, which can lead to decreased performance or wasted energy during low-temperature fast charging. For the first... For any random scenario, if the battery temperature at all time steps within the prediction time domain falls within the safe range, then the scenario is considered a safe scenario; otherwise, it is considered an unsafe scenario. The safety probability obtained based on scenario statistics can be expressed as:

[0075]

[0076] in, This is the safety indicator value corresponding to the j-th random scenario. When the battery temperature is within the preset safety range for all time steps in the prediction time domain under the j-th scenario... =1; When the battery temperature exceeds the preset safety range at any time step, =0. This probability is not calculated based on a single worst-case or average condition, but rather based on statistics from multiple random scenarios. Therefore, it can more realistically reflect the degree of thermal safety satisfaction under future uncertain conditions.

[0077] Precooling energy consumption can be determined based on the cumulative amount of precooling power within the predicted time domain, and its expected form can be expressed as:

[0078]

[0079] in, Represents the parameter vector The corresponding precooling energy consumption, Let represent the precooling power at time step t, and From the parameter vector The precooling start-up time and the control node of the precooling power curve are determined. This indicates the time step. Since the pre-cooling power curve is determined by the same parameter vector, the differences under different random scenarios are mainly reflected in the battery load power, the external ambient temperature, and the battery temperature trajectory derived from them.

[0080] Preferably, when considering the influence of ambient temperature on cooling efficiency, the above energy consumption can be corrected according to the coefficient of performance (COP).

[0081] like Figure 6 As shown, in step S35, based on the expected value of the pre-cooling energy consumption and the probability that the battery temperature is within the preset safe range, a chance-constrained programming problem is constructed:

[0082]

[0083] in, The confidence level is pre-set. The planning problem aims to find a set of pre-cooling start-up times and pre-cooling power curve parameters among a large number of possible future operating scenarios, minimizing the expected pre-cooling energy consumption while ensuring that the battery temperature remains within a preset safe range for all time steps in the prediction time domain, with a probability not lower than the pre-set confidence level. In other words, this method constrains the entire temperature trajectory probabilistically, thus avoiding energy waste due to excessive conservatism and insufficient pre-cooling and thermal safety risks due to excessive optimism.

[0084] like Figure 7As shown, in step S4, the opportunity-constrained planning problem is solved to obtain a precooling path planning scheme that includes the precooling start time and the precooling power target curve.

[0085] Specifically, an iterative optimization algorithm based on random scenario sampling can be used to solve the problem. First, an initial parameter vector is set, for example, the pre-cooling start time is set to a certain number of time steps before the expected entry into a high-heat-risk section, and the pre-cooling power is set to medium power. Then, in each iteration, Monte Carlo sampling is re-executed to update the scenario set, and the expected pre-cooling energy consumption and the probability that the battery temperature is within the preset safe range are recalculated based on the updated scenario set. If the current parameter vector does not meet the safe probability constraint, the pre-cooling start time can be advanced, the pre-cooling power increased, or the pre-cooling duration extended; if the current parameter vector meets the safe probability constraint but energy consumption can still be reduced, the pre-cooling start time can be delayed, the pre-cooling power reduced in some time steps, or the pre-cooling duration shortened.

[0086] The iterative optimization algorithm may employ stochastic gradient descent, particle swarm optimization, covariance matrix adaptive evolution strategy, Bayesian optimization, cross-entropy method, or sequential quadratic programming.

[0087] When using particle swarm optimization (PSO), the parameter vector can be used as the particle position, and the expected precooling energy consumption and constraint violation penalty term can be combined to form the fitness function, allowing the particle swarm to gradually converge towards the feasible region that satisfies the chance constraints during the search process. For optimization problems that are non-convex, non-smooth, or involve characteristic functions, this type of heuristic algorithm can improve the stability of the solution.

[0088] The purpose of re-performing Monte Carlo sampling in each iteration is to reduce the accidental impact of a single sample set on the optimization results, avoid over-fitting the pre-cooled path planning scheme to a certain set of random samples, and thus improve the robustness of the obtained scheme in real uncertainty environments.

[0089] Preferably, a maximum number of iterations, a maximum computation time, a threshold for objective function changes, and a condition for maintaining the safety probability can be set. When the change in pre-cooling energy consumption over multiple consecutive iterations is less than a preset threshold, and the safety probability consistently meets a preset confidence level, the iteration stops, and the optimal parameter vector is obtained. Alternatively, when the number of iterations reaches a preset maximum number of iterations, or the planning computation time reaches the maximum computation time allowed by the vehicle controller, the iteration stops, and the parameter vector with the lowest pre-cooling energy consumption is selected as the output from the candidate solutions that currently satisfy the chance constraints.

[0090] After the solution is obtained, the optimal parameter vector is restored to the precooling start-up time and the precooling power target curve. Specifically, the precooling start-up time is extracted from the optimal parameter vector, and the precooling power values ​​of each control segment are expanded according to the corresponding time steps to form a complete time-power mapping relationship. This precooling power target curve can be issued to the battery precooling system as a formatted control command sequence.

[0091] Preferably, before the power target curve is issued, it can be limited and smoothed to reduce sudden changes in actuator control, but this process does not change the overall safety probability constraints and energy consumption optimization objectives of the precooling path planning scheme.

[0092] In step S5, the battery precooling system is controlled to perform active precooling on the battery according to the precooling path planning scheme.

[0093] Specifically, the system monitors the current time or location in real time. When the vehicle reaches the planned pre-cooling start time, or when it reaches the planned time window for high-heat-risk road sections, fast-charging stations, or long uphill sections, a real-time pre-cooling power command is generated based on the pre-cooling power target curve and sent to the actuators of the battery pre-cooling system via the vehicle bus. For the compressor, the refrigerant mass flow rate can be controlled by adjusting the compressor speed; for the electronic water pump, the coolant circulation flow rate can be controlled by adjusting the water pump speed; for the electronic expansion valve, the refrigerant throttling state can be controlled by adjusting the valve opening; and for the three-way valve or reversing valve, the distribution ratio of coolant between the battery branch and other thermal management branches can be controlled by adjusting the valve position.

[0094] Preferably, based on pre-calibrated mapping relationships between compressor speed and cooling power, pump speed and flow rate, and valve opening and differential pressure, the precooling power command is converted into specific control signals for each actuator. Through this mapping relationship, the precooling power target curve output by the planning layer can be implemented as an executable control quantity at the lower level.

[0095] During the pre-cooling process, real-time temperature data is acquired from multiple temperature measurement points within the battery pack. These temperature sensors can be positioned near the coolant inlet and outlet, the center of the battery modules, and areas prone to heat concentration. The highest temperature measurement point can be used as a safety feedback value, while the average temperature can be used as a reference value for energy consumption adjustment. When the highest temperature measurement point approaches the safety limit, the pre-cooling capacity is prioritized for increase; when the temperature distribution is relatively uniform and below the target temperature, the pre-cooling power is appropriately reduced. This feedback strategy improves the ability to identify local hotspots and avoids masking local thermal risks by relying solely on average temperature control.

[0096] like Figure 8As shown, the present invention also includes step S6, which is to collect the actual load power and actual ambient temperature in real time during the actual driving process of the vehicle; when the actual load power or actual ambient temperature falls outside the preset confidence interval of the corresponding time step determined according to the corresponding probability distribution sequence, the currently collected actual driving data is used to update the trip feature vector and the current state information, and steps S2 to S4 are retried to generate and execute the updated pre-cooling path planning scheme.

[0097] Specifically, during vehicle operation, the system continuously collects data on actual battery load power, actual ambient temperature, current location, current vehicle speed, current battery temperature, and current state of charge. The actual load power and ambient temperature are then compared with the corresponding time-step probability distribution output in step S2. If the probability distribution is Gaussian, the preset confidence interval can be a range of several standard deviations plus or minus the mean; for example, a range of approximately 95% confidence level corresponds to two standard deviations plus or minus the mean.

[0098] It should be noted that the confidence interval here is used to determine whether the actual operating conditions have deviated from the expected range of the probabilistic prediction model, thereby deciding whether to trigger closed-loop reprogramming; it is different from the pre-set confidence in the chance-constrained programming problem, which is used to constrain the lower probability limit of the battery temperature being within the preset safe range in the prediction time domain.

[0099] When the actual load power or actual ambient temperature exceeds the corresponding confidence interval, it indicates that the current actual driving conditions have significantly deviated from the original predicted distribution, and the input basis of the original pre-cooling path planning scheme has changed, triggering replanning. During replanning, the trip feature vector is reconstructed using the actual driving data already collected by the vehicle, the current battery status, and the remaining navigation path, and probability prediction, scene sampling, and chance constraint planning are re-executed.

[0100] Therefore, this method forms a closed-loop mechanism of rolling prediction, rolling planning, and rolling execution. When there is sudden traffic congestion, vehicle detours, rapid increase in outside temperature, continuous high-power acceleration by drivers, or changes in fast charging demand, the pre-cooling start-up time and pre-cooling power curve can be adjusted in a timely manner to prevent the original plan from becoming invalid after changes in actual operating conditions.

[0101] Preferably, to avoid frequent replanning caused by short-term sensor noise, trigger hold conditions and a minimum replanning interval can be set. For example, replanning is only triggered when the actual load power or actual ambient temperature falls outside the confidence interval for multiple consecutive sampling periods; a minimum time interval of 30 to 300 seconds is set between two adjacent replanning sessions. This preferred strategy can reduce the computational burden and execution fluctuations caused by frequent switching while maintaining the planning adaptive capability.

[0102] like Figure 9As shown, in another embodiment, the pre-set confidence level in the opportunity-constrained programming problem increases with the thermal risk level of future journeys. The thermal risk level can be determined based on at least one of fast charging demand, long uphill sections, and hot, congested sections.

[0103] For example, if the navigation destination is a fast charging station and the expected state of charge upon arrival is below a preset threshold, a fast charging requirement is identified, and the thermal risk level is increased. Since fast charging inputs a large amount of electrical energy into the battery in a short time, significantly increasing battery heat generation, more robust pre-cooling of the battery before reaching the fast charging station is practically meaningful. In this case, the pre-set confidence level in the opportunity-constrained programming problem can be increased from the default value to a more conservative level, such as from 90% to 95% or 99%.

[0104] For example, when the future path includes a long, continuous uphill section with a steep gradient, the vehicle needs to maintain high driving power for an extended period, increasing the battery discharge load and consequently, the heat generated by the battery. In this case, the thermal risk level can be increased, requiring the opportunity-constrained programming problem to adopt a higher safety probability requirement. Another example is when the future path anticipates hot and congested road sections. The vehicle speed decreases, the oncoming airflow weakens, and the battery pack's natural heat dissipation capacity declines. Simultaneously, the high ambient temperature reduces the temperature difference between the battery and the environment, making the battery temperature rise more rapidly. In such scenarios, increasing the pre-set confidence level provides a greater margin for battery thermal safety.

[0105] Thermal risk levels can be categorized into low, medium, and high, each corresponding to different pre-set reliability levels. The assessment of thermal risk levels can be achieved using lookup table rules, fuzzy logic inference rules, weighted scoring models, or machine learning classification models. For example, a thermal risk score can be calculated based on projected fast charging demand, continuous uphill length, average slope, ambient temperature, congestion index, current battery temperature, and current state of charge, and the corresponding pre-set reliability level can be determined based on the interval in which the thermal risk score falls.

[0106] This adaptive confidence level mechanism allows the conservatism of the pre-cooling strategy to match the actual trip risk. In low-risk trips, a lower confidence level can be used to reduce over-pre-cooling and range loss; in high-risk trips, a higher confidence level can be used to increase the probability of the battery temperature remaining within the safe range. Compared to a fixed safety margin approach, this method can dynamically adjust thermal safety requirements based on actual trip characteristics, achieving a more refined balance between safety and energy efficiency.

[0107] Preferably, the thermal risk level can also be adjusted based on battery health status, vehicle load, passenger compartment air conditioning requirements, charging station queuing time, and battery fast charging temperature window. For example, when the battery health status is low or the battery internal resistance increases, the thermal risk level is increased; when the passenger compartment air conditioning requirements are high and the cooling capacity needs to be distributed between the passenger compartment and the battery, the maximum power available for battery pre-cooling is dynamically reduced; when the expected queuing time at the fast charging station is long, the pre-cooling start can be appropriately delayed to avoid premature cooling before arrival and subsequent temperature rise.

[0108] In summary, this method first acquires future travel information and the current battery state, forming a travel feature vector. Then, a probabilistic prediction model outputs a probability distribution sequence of battery load power and external ambient temperature in the future prediction time domain. Next, Monte Carlo sampling transforms the probability distribution into multiple random scenarios. Subsequently, a battery thermal balance recursive model is used to calculate the battery temperature trajectory under each random scenario. Furthermore, a chance-constrained planning problem is constructed based on the scenario safety ratio and the expected pre-cooling energy consumption. Finally, iterative optimization yields the pre-cooling start time and the pre-cooling power target curve, which are then converted into control signals executable by actuators such as compressors, pumps, and valves. This data flow connects layer by layer from information acquisition, probabilistic prediction, random scenario generation, thermal state recursion, chance-constrained optimization to execution control. The output of each layer serves as the input of the next layer, forming a logically closed active pre-cooling path planning method.

[0109] This method explicitly models the uncertainties of future battery load power and external ambient temperature as a probability distribution sequence, and constructs a chance-constrained programming problem based on this probability distribution sequence. Through the synergy between the probabilistic prediction model, random scenario sampling, battery thermal balance recursive model, and chance-constrained optimization, the expected value of pre-cooling energy consumption can be minimized while meeting the pre-set confidence thermal safety requirements.

[0110] This method enables precooling control to no longer rely on fixed conservative margins or a single ideal operating condition, but to make decisions based on quantifiable safety probabilities and calculable energy consumption targets. It can effectively improve the problem that existing deterministic precooling planning is unable to balance battery thermal safety and energy consumption economy under uncertain future operating conditions.

[0111] Furthermore, this method can determine whether the actual operating conditions have deviated from the original predicted distribution, rather than relying solely on fixed time intervals or fixed deviation thresholds for updates. This mechanism enables the pre-cooling strategy to have rolling correction capabilities, allowing for timely updates to the pre-cooling start time and pre-cooling power target curve when vehicles change lanes, traffic conditions change abruptly, ambient temperature changes abnormally, or battery load continues to rise, thereby improving the robustness of active pre-cooling control.

[0112] In summary, the beneficial effects of the present invention include at least the following:

[0113] By acquiring navigation path information, traffic situation information, ambient temperature information, and current battery status information for the vehicle's future journey, and based on this information, performing uncertainty predictions on battery load power and external ambient temperature in the future prediction time domain, the uncertainties in future operating conditions caused by factors such as changes in vehicle speed, road congestion, driving behavior, and weather changes during actual driving can be incorporated into the battery active pre-cooling path planning process.

[0114] Based on the uncertain predicted distribution sequence of future battery load power and external ambient temperature, a chance-constrained programming problem is constructed in conjunction with the battery thermal state model. The constraint condition is that the probability of the battery temperature being within a preset safe range in the predicted time domain is not lower than a preset confidence level. Therefore, the pre-cooling path planning no longer makes decisions only for a certain ideal or conservative operating condition, but determines the pre-cooling start time and the pre-cooling power target curve under the premise of meeting the preset thermal safety probability requirements. This can reduce the risk of insufficient pre-cooling caused by deviations in future operating conditions and improve the thermal safety margin of the power battery in scenarios such as high-temperature environment, high-power discharge, and fast charging pre-cooling.

[0115] With the goal of minimizing the expected value of precooling energy consumption, the timing of precooling initiation and the precooling power are optimized in a coordinated manner while meeting the battery thermal safety probability constraints. This avoids the energy waste caused by premature precooling initiation or continuous use of high precooling power under existing conservative prediction strategies, reduces the adverse impact of the battery thermal management system on the vehicle's driving range, and improves the energy efficiency of active precooling control.

[0116] By acquiring actual load power and ambient temperature during actual driving, and triggering replanning when the actual operating conditions deviate from the confidence interval determined by the corresponding probability distribution sequence, the pre-cooling path planning scheme can be rolled over and corrected based on vehicle detours, sudden changes in traffic conditions, changes in ambient temperature, or changes in driving behavior. This improves the adaptability of pre-cooling path planning to dynamic driving environments and reduces the impact of one-time prediction errors on the pre-cooling control effect.

[0117] The thermal risk level can be determined based on factors such as fast charging demand, long uphill sections, and high-temperature congestion sections during future journeys, and the pre-set confidence level in the opportunity constraint planning can be adjusted accordingly. As a result, the thermal safety constraint strength can be increased in high thermal risk scenarios, and the conservatism can be appropriately reduced in low thermal risk scenarios, thereby achieving an adaptive balance between battery thermal safety and pre-cooling energy consumption.

[0118] like Figure 10As shown, to implement the above-mentioned active battery precooling path planning method, the present invention also provides a thermal management system. This thermal management system is used to execute the aforementioned active battery precooling path planning method and control the battery precooling system to actively precool the power battery according to the generated precooling path planning scheme.

[0119] The thermal management system includes an information acquisition module, a feature processing module, a prediction and planning module, and a precooling execution control module.

[0120] The information acquisition module is used to acquire trip-related information for the vehicle's future journey and the current battery status information. The trip-related information may include navigation route information, traffic situation information, and ambient temperature information, while the current battery status information may include the current battery temperature and current state of charge. The information acquisition module can communicate with the in-vehicle navigation module, communication module, battery management system, and vehicle controller to obtain the input data required to execute the aforementioned methods.

[0121] The feature processing module processes the trip-related information to obtain a trip feature vector. Specifically, the feature processing module can divide the future trip into multiple prediction units based on the navigation path and extract features such as road speed limits, road segment lengths, gradients, congestion indices, estimated travel times, and ambient temperatures corresponding to each prediction unit. Preferably, the feature processing module can also perform timestamp alignment, outlier removal, filtering, or road segment matching on data from different sources to improve the consistency of the input data.

[0122] The predictive planning module is used to execute the aforementioned probabilistic prediction and chance-constrained planning processes based on the travel feature vector and the current battery state information to obtain a pre-cooling path planning scheme. The pre-cooling path planning scheme includes the pre-cooling start time and the pre-cooling power target curve. Specifically, the predictive planning module can call a pre-built probabilistic prediction model to obtain the uncertainty prediction distribution sequence of battery load power and external ambient temperature in the future prediction time domain, and combine it with the battery thermal state model to construct a chance-constrained planning problem to solve for a pre-cooling path planning scheme that meets battery thermal safety requirements and has low pre-cooling energy consumption. The probabilistic prediction model, battery thermal state model, chance-constrained planning problem, and solution process can all adopt the corresponding implementation methods in the aforementioned active battery pre-cooling path planning method, and will not be repeated here.

[0123] The precooling execution control module is used to control the battery precooling system to perform active precooling according to the precooling path planning scheme output by the prediction planning module. Specifically, the precooling execution control module can generate precooling control commands based on the precooling start time and the precooling power target curve, and convert the precooling control commands into actuator control quantities such as compressor speed, electric water pump speed, electronic expansion valve opening, valve opening, or coolant target temperature.

[0124] In one optional embodiment, the battery precooling system may include a compressor refrigeration circuit, a coolant circulation circuit, a heat exchanger, an electric water pump, an electronic expansion valve, a liquid cooling plate, and a temperature sensor. The compressor refrigeration circuit provides cooling, the coolant circulation circuit delivers the cooling to the power battery, the liquid cooling plate exchanges heat with the power battery, and the temperature sensor collects the temperature information of the power battery. The precooling execution control module can convert the precooling power target curve into corresponding actuator control signals according to a preset actuator calibration relationship.

[0125] Preferably, the thermal management system further includes a closed-loop update module. This module acquires the actual load power, actual ambient temperature, and actual battery temperature during actual vehicle operation. When the actual operating conditions deviate from the aforementioned probability prediction results, it triggers the prediction planning module to regenerate the pre-cooling path planning scheme. Specifically, when the actual load power or actual ambient temperature falls outside a preset confidence interval for the corresponding time step determined according to the corresponding probability distribution sequence, the closed-loop update module can update the input information using the currently collected actual driving data and trigger replanning. Thus, the pre-cooling strategy can be continuously corrected when the vehicle changes lanes, traffic conditions change, ambient temperature changes, or driving conditions change.

[0126] Preferably, the thermal management system may further include a risk level determination module. This module determines the thermal risk level based on at least one of the following factors during the future trip: fast charging demand, long uphill sections, and high-temperature congested sections. It then adjusts the pre-set confidence level in the opportunity-constrained programming based on the thermal risk level. A higher thermal risk level corresponds to a higher pre-set confidence level, and vice versa.

[0127] In practical deployment, the information acquisition module, feature processing module, prediction planning module, pre-cooling execution control module, closed-loop update module, and risk level determination module can be implemented as software programs, hardware circuits, or a combination of both. These modules can be integrated into the battery thermal management controller, vehicle controller, vehicle domain controller, or central computing platform, or they can be implemented collaboratively by the vehicle-side controller and a cloud server. For example, training of the probabilistic prediction model can be completed in the cloud, while model inference and pre-cooling execution control can be completed on the vehicle side; alternatively, both probabilistic prediction and chance-constrained planning can be completed on the vehicle side.

[0128] The aforementioned thermal management system can provide data acquisition, feature processing, prediction planning, and execution control conditions for the aforementioned active battery precooling path planning method, enabling the method to be implemented in vehicle thermal management scenarios.

[0129] 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 invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0130] 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 battery active precooling path planning method, characterized in that, Includes the following steps: S1. Obtain trip-related information for the vehicle's future journey, extract features from the trip-related information to obtain a trip feature vector; simultaneously, obtain the current state information of the battery; S2. Input the trip feature vector and current state information into a pre-built probability prediction model. The probability prediction model outputs the probability distribution sequence of battery load power and the probability distribution sequence of external ambient temperature in the future prediction time domain, as the uncertainty prediction distribution sequence. S3. Using the uncertainty prediction distribution sequence as input, and combining it with the preset battery thermal state model, construct a chance-constrained programming problem. The chance-constrained programming problem takes the pre-cooling start-up time and the pre-cooling power curve as decision variables, takes the minimum expected value of pre-cooling energy consumption as the optimization objective, and takes the probability that the battery temperature is within the preset safe range for all time steps in the prediction time domain under the random scenario represented by the uncertainty prediction distribution sequence is not less than the preset confidence level as the constraint condition. S4. Solve the opportunity-constrained programming problem to obtain a precooling path planning scheme that includes the precooling start time and the precooling power target curve; S5. According to the precooling path planning scheme, control the battery precooling system to perform active precooling on the battery.

2. The battery active precooling path planning method according to claim 1, characterized in that, In S1, the trip-related information includes navigation path information, traffic situation information, and ambient temperature information; the current status information includes the current battery temperature and the current state of charge.

3. The battery active precooling path planning method according to claim 1, characterized in that, In S2, the probability prediction model is a sequence-to-sequence architecture-based probability prediction model, which includes an encoding module and a probability decoding module. The step of inputting the trip feature vector and current state information into a pre-constructed probability prediction model, and having the probability prediction model output the probability distribution sequence of battery load power and the probability distribution sequence of external ambient temperature in the future prediction time domain, includes: The encoding module is used to encode the features of each road segment and the current state information in the trip feature vector to obtain the temporal context features; Based on temporal context features, the probability decoding module outputs the distribution parameters of battery load power and external ambient temperature at each time step in the prediction time domain, so as to form the probability distribution sequence of battery load power and the probability distribution sequence of external ambient temperature, respectively.

4. The battery active precooling path planning method according to claim 1, characterized in that, In S3, the battery thermal state model is a battery thermal balance recursive model, which is used to calculate the corresponding battery temperature change based on the battery load power, external ambient temperature and pre-cooling power at each time step, so as to recursively obtain the battery temperature at each time step.

5. The battery active precooling path planning method according to claim 4, characterized in that, S3 specifically includes: S31. The precooling start-up time and the precooling power curve are parameterized into a parameter vector; S32. Based on the probability distribution sequence, generate a set of scenarios representing multiple possible future working conditions through Monte Carlo sampling; S33. For each scenario in the scenario set, using the battery thermal balance recursive model, calculate the battery temperature at each time step in the prediction time domain under that scenario based on the battery load power time series, the external ambient temperature time series, and the pre-cooling start-up time and pre-cooling power curve represented by the parameter vector. S34. The average pre-cooling energy consumption under the scenario set is used to construct the expected value of the pre-cooling energy consumption; the proportion of the number of scenarios in which the battery temperature is within the preset safe range at all time steps in the prediction time domain to the total number of scenarios in the scenario set is used to construct the probability that the battery temperature is within the preset safe range. S35. Based on the expected value of the pre-cooling energy consumption and the probability that the battery temperature is within a preset safe range, construct the opportunity-constrained programming problem.

6. The battery active precooling path planning method according to claim 5, characterized in that, In S4, the opportunity-constrained programming problem is solved to obtain a precooling path planning scheme that includes the precooling start-up time and the precooling power target curve, including: An iterative optimization algorithm based on random scenario sampling is used to solve the chance-constrained planning problem. In each iteration, the Monte Carlo sampling is re-executed to update the scene set, and the expected value of the precooling energy consumption and the probability that the battery temperature is within the preset safe range are recalculated based on the updated scene set to update the parameter vector until the constraint conditions are met and the expected value of the precooling energy consumption converges to obtain the optimal parameter vector. The optimal parameter vector is then restored to the pre-cooling start-up time and pre-cooling power target curve.

7. The battery active precooling path planning method according to claim 4, characterized in that, The battery thermal balance recursive model is characterized by the following equation: in, Let be the battery temperature at time step t. Let be the external ambient temperature at time step t. Let be the battery load power at time step t. Let be the precooling power at time step t. For time step, The equivalent heat capacity of the battery. Let the heat generated by the battery load power be the function. This is used to characterize the heat dissipation power function of the precooling system at the current battery temperature and ambient temperature.

8. The battery active precooling path planning method according to claim 1, characterized in that, It also includes S6: Real-time collection of actual load power and actual ambient temperature during actual vehicle operation; When the actual load power or actual ambient temperature falls outside the preset information interval of the corresponding time step determined according to the corresponding probability distribution sequence, the travel feature vector and current state information are updated using the currently collected actual driving data, and steps S2 to S4 are retried to generate and execute the updated pre-cooling path planning scheme.

9. The battery active precooling path planning method according to claim 1, characterized in that, The pre-set confidence level increases as the heat risk level of future trips increases; The thermal risk level is determined based on at least one of the following: fast charging demand, long uphill road sections, and high-temperature congested road sections.

10. A thermal management system for implementing an active pre-cooling path planning method for batteries, characterized in that, include: The information acquisition module is used to acquire trip-related information for the vehicle's future journey and the current battery status information. The trip-related information includes navigation path information, traffic situation information, and ambient temperature information. The current battery status information includes the current battery temperature and the current state of charge. The feature processing module, connected to the information acquisition module, is used to process the trip-related information to obtain a trip feature vector; The prediction planning module, connected to the feature processing module, is used to obtain the uncertainty prediction distribution sequence of battery load power and external ambient temperature in the future prediction time domain based on the travel feature vector and the current state information of the battery, and to construct an opportunity-constrained planning problem based on the uncertainty prediction distribution sequence and the battery thermal state model to solve the precooling path planning scheme, which includes the precooling start time and the precooling power target curve. The precooling execution control module, connected to the prediction planning module, is used to generate precooling control instructions based on the precooling path planning scheme, and to control the battery precooling system to perform active precooling on the power battery based on the precooling control instructions.

Images