Flow regulation method for power battery liquid cooling system based on working condition prediction

Abstract

The application discloses a power battery liquid cooling system flow regulation method based on working condition prediction, relates to the technical field of battery cooling control, and comprises the following steps: a controller collects the module temperature, the cooling liquid inlet and outlet temperature, the current state of charge and the vehicle working condition, calculates and predicts the temperature and the temperature rise trend based on a lightweight thermal model, calculates the thermal margin and divides the urgent area and the non-urgent area, and outputs the temperature upper limit constraint, the temperature difference constraint, the pump speed boundary and the hotspot risk weight. When the thermal margin is sufficient and the temperature rise trend is gentle, the calibration window is entered, the limited disturbance is applied to the water pump speed and the branch valve opening, the equivalent resistance parameter and the heat exchange marginal slope are updated according to the disturbance-response, and the reliability and the abnormal flag are given. Then, the valve position is distributed according to the change of the hotspot temperature corresponding to the water pump power, the pump speed is determined in the high-efficiency working area of the water pump, and the abnormality is retreated to the conservative control, so that the pump energy consumption is reduced and the adaptability to assembly difference and aging drift is improved.

Description

Technical Field

[0001] This invention relates to the field of battery cooling control technology, specifically to a method for regulating the flow rate of a power battery liquid cooling system based on operating condition prediction. Background Technology

[0002] With the popularization of new energy vehicles, the power battery, as the energy and power source of the entire vehicle, directly affects the reliability and user experience of the vehicle due to its thermal safety and lifespan. To meet the heat dissipation requirements of high-energy-density batteries under conditions such as high ambient temperature, prolonged low-speed congestion, high-speed cruising, continuous hill climbing, and high-frequency acceleration, vehicles are mostly equipped with liquid-cooled circuits using coolant as the medium. These circuits conduct heat from the battery cells or modules to the radiator or cooler through adjustable-speed water pumps, branch valves, and cold plates. In practical applications, battery heating varies greatly with current and internal resistance, and is significantly affected by state of charge, aging, coolant temperature, and viscosity. The parallel branches and manifolds within the battery pack form a hydraulic network, and the voltage drop in the branches is coupled with the valve opening, meaning that the flow rate received by each module may differ under the same total flow rate, posing risks of hot spots and temperature differences. Furthermore, pump power consumption is also closely related to system voltage drop. When using conservative control margin calibration methods, pump power can lead to unpredictable vehicle energy consumption and range, and may also cause noise and component wear.

[0003] Existing liquid cooling flow rate regulation technologies are mostly based on temperature feedback threshold control, proportional-integral control, lookup table mapping, or rule-based control. Some schemes also use short-term operating conditions or heat load predictions as a basis for adjusting pump speed and valve position in advance to meet maximum temperature and temperature difference constraints. However, these methods lack pre-set parameters, fixed branch resistance and valve characteristic assumptions, and characterization of pump efficiency variations with operating points, hydraulic network coupling, and parameter drift during engineering implementation. Due to factors such as manufacturing tolerances, uneven assembly, air bubble entrainment, coolant aging, or temperature changes, branch resistance and valve characteristics may deviate, causing the actual flow rate and pressure drop corresponding to the same control command to deviate from the calibrated operating conditions. In such cases, the control method needs to increase safety margins or frequently increase pump speed to avoid overheating, thus increasing pump energy consumption and making flow distribution uneconomical. Furthermore, rapid changes in operating conditions may lead to temperature fluctuations and increased duration of local hot spots, resulting in power derating, accelerated decrease in battery cell consistency, or reduced component lifespan.

[0004] Therefore, given the rapid changes in vehicle operating conditions and heat load, and the drifting of the hydraulic characteristics of the liquid cooling circuit due to manufacturing differences and usage conditions, existing flow regulation methods struggle to achieve sustainable and economical pump power consumption and flow distribution while ensuring that the maximum temperature and temperature difference of the power battery are controlled. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] To address the shortcomings of existing technologies, this invention provides a flow regulation method for a power battery liquid cooling system based on operating condition prediction. This method employs output temperature upper limit constraints, temperature difference constraints, pump speed boundaries, and hot spot risk weights. When the thermal margin is sufficient and the temperature rise trend is gradual, a calibration window is entered. Constrained perturbations are applied to the pump speed and branch valve opening. The equivalent resistance parameters and heat transfer marginal slope are updated based on the perturbation-response relationship, and confidence and anomaly indicators are provided. Subsequently, valve positions are allocated according to the hot spot temperature changes corresponding to pump power, and the pump speed is determined within the pump's high-efficiency operating range. In case of anomalies, conservative control is applied, thereby reducing pump energy consumption and improving adaptability to assembly differences, thus solving the technical problems described in the background art.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] The flow regulation method for power battery liquid cooling system based on operating condition prediction includes: the controller collects module temperature, coolant inlet and outlet temperature, current and state of charge and vehicle operating condition information; the controller calculates and predicts temperature trajectory based on lightweight thermal model and obtains predicted temperature rise trend; the controller calculates thermal margin and divides the urgent zone and non-urgent zone; and the controller outputs temperature upper limit constraint, temperature difference constraint and hot spot risk weight.

[0010] When the heat margin meets the preset threshold and the predicted temperature rise trend is gradual, a limited disturbance is applied to the pump speed and branch valve opening in the non-critical zone, and the battery temperature and coolant temperature difference are monitored. The equivalent resistance parameters and heat transfer margin slope are updated based on the disturbance-response.

[0011] The controller, based on hot spot risk weights, upper temperature limit constraints, temperature difference constraints, and combined with updated equivalent resistance parameters and heat transfer marginal slope, first allocates the branch valve opening and then determines the pump speed. The allocation metric is the change in hot spot temperature corresponding to the unit pump work, while also satisfying the pump's high-efficiency operating range constraints.

[0012] Furthermore, ambient temperature and water pump speed electrical parameters are collected, and the module temperature and coolant inlet and outlet temperatures are time-aligned and validated. The aligned signals are written into the prediction input record, and abnormal signals are replaced with the most recent valid value and written into the diagnostic record. The minimum safe pump speed boundary is then output.

[0013] Furthermore, the heat margin is obtained by performing logarithmic and exponential aggregation operations on the predicted temperature trajectory within the prediction window, and negative heat margins are truncated; the predicted temperature rise trend is obtained by the temperature difference between adjacent prediction steps, and the urgent zone and the non-urgent zone are jointly determined by the heat margin and the predicted temperature rise trend.

[0014] Furthermore, the lightweight thermal model uses discrete recursion with the control cycle as the step size to generate the predicted temperature trajectory. The heat generation is calculated from the current and the equivalent internal resistance. The equivalent internal resistance is obtained by bilinear interpolation from a two-dimensional table of state of charge and module temperature index. The consistency of the recursive boundary is checked by the coolant outlet temperature.

[0015] Furthermore, the calibration window is triggered by the thermal margin meeting the preset threshold and the predicted temperature rise trend meeting the condition of being gradual. After triggering, the branch with the smaller hot spot risk weight is selected from the non-urgent area as the excited branch, and the branch sharing the manifold with it is set as the observation branch, and the pump speed is not lower than the minimum safe pump speed boundary.

[0016] Furthermore, the restricted disturbance is applied in a time-sharing sequence. First, while keeping the pump speed constant, a step change, hold, and recovery are performed on the opening of a single branch valve. Then, while keeping the branch valve opening constant, a step change, hold, and recovery are performed on the pump speed. The step change amplitude is taken as an integer multiple of the actuator's minimum resolvable step and is constrained by the valve action budget.

[0017] Furthermore, the disturbance-response consists of calibration observation records based on the module temperature, coolant inlet and outlet temperatures, coolant temperature difference, and water pump speed electrical parameters within the observation segment. The residuals are synthesized from the coolant outlet temperature change residuals and the hot spot temperature change slope residuals according to preset weights, and a model credibility index is generated accordingly.

[0018] Furthermore, the updated parameters also include the valve opening-flow relationship deviation. The equivalent resistance parameter and the valve opening-flow relationship deviation are limited to a preset physically feasible range through projection mapping. Anomaly flags are calculated based on the model credibility index and the perturbation-response consistency. When the anomaly flag is in a valid state, the calibration window is terminated and the update is frozen.

[0019] Furthermore, the allocation of branch valve openings is based on the hot spot temperature change corresponding to the unit pump work, and the accessibility is checked by combining the hot spot risk weight, heat exchange marginal slope, equivalent resistance parameter and valve opening-flow relationship deviation; the branch valve opening corresponding to the urgent zone is prohibited from being reduced, and the branch valve opening is adjusted in a monotonic step manner and meets the valve action budget.

[0020] Furthermore, the high-efficiency operating range constraint of the water pump is achieved by looking up the water pump efficiency table. The water pump efficiency is obtained by interpolation using the water pump speed and system pressure difference as indexes. The system pressure difference is collected by the differential pressure sensor and estimated by looking up the water pump speed electrical parameter information when the differential pressure sensor is unavailable. The water pump speed is checked step by step from the minimum safe pump speed boundary. If the check fails, the branch valve opening is adjusted first and then the water pump speed is adjusted.

[0021] (III) Beneficial Effects

[0022] This invention provides a flow rate regulation method for a power battery liquid cooling system based on operating condition prediction, which has the following beneficial effects:

[0023] In each control cycle, module temperature, coolant inlet and outlet temperatures, current, state of charge, and vehicle operating conditions are collected. The predicted temperature trajectory and temperature rise trend are obtained through a lightweight thermal model, enabling control to shift from current temperature feedback to predicted hot spot constraints, avoiding overheating and temperature difference fluctuations caused by lag adjustment. Based on the predicted temperature trajectory, thermal margin and urgent and non-urgent zones are calculated, resulting in upper temperature limit constraints, temperature difference constraints, minimum safe pump speed boundaries, and hot spot risk weights. Subsequent calibration and allocation determine priorities within a unified boundary to avoid different decisions when operating conditions change abruptly.

[0024] When the thermal margin meets the threshold and the temperature rise trend is gradual, the pump speed and branch valve opening are subject to limited disturbances. The equivalent resistance parameters and heat transfer marginal slope are updated according to the response of the disturbances, so that the loop model is self-corrected by assembly differences, air resistance and aging drift, reducing the accumulation of deviations caused by fixed calibration maps.

[0025] The calibration window calculates the confidence level and generates an anomaly flag. If the confidence level is insufficient or an anomaly occurs, the calibration is immediately stopped and conservative control is rolled back. Parameter updates have acceptance conditions and lockout logic to prevent unreliable parameters from entering the valve position allocation and pump speed determination process, thereby enhancing controllability under abnormal conditions.

[0026] Under the constraints of hot spot risk weight, upper temperature limit, and temperature difference, the branch valve opening is allocated and the pump speed is determined by the equivalent resistance parameter and the heat exchange marginal slope. The hot spot temperature change corresponding to a unit pump work is used as the metric. Cooling resources are concentrated in branches with more significant marginal cooling and lower hydraulic cost, thereby reducing ineffective flow supply.

[0027] When determining the pump speed, a constraint on the pump's high-efficiency operating range is added. When it is difficult to balance all constraints, the opening of the branch valve is finely adjusted first to change the system resistance before adjusting the pump speed. A fallback path is retained. The pump and valve can operate continuously in both reliable and unreliable parameter states and meet the upper temperature limit and temperature difference constraints. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the flow regulation method for a power battery liquid cooling system based on operating condition prediction according to the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Please see Figure 1 This invention provides a flow rate regulation method for a power battery liquid cooling system based on operating condition prediction, including:

[0031] Step 1: The available temperature, current, state of charge, and driving status information are sequentially mapped to the hotspot temperature change results within the prediction window to form thermal margin stratification, control objectives, and hotspot risk weights, enabling calibration and flow decision-making under constraint boundaries and risk characterization in subsequent steps.

[0032] The temperature changes of power batteries exhibit thermal inertia, and the heat transfer response of the cooling circuit is lag-dependent. Adjusting solely based on the current module temperature can easily lead to lag compensation or a tendency to conservatively increase the pump speed. Meanwhile, changes in battery current and state of charge alter the heat generation intensity, and changes in driving conditions can cause rapid shifts in future heat load within a short period. Therefore, it is necessary to organize multi-source observations into a consistent predictive input within the control cycle, and then recursively derive the hot spot temperature sequence within the prediction window using a lightweight thermal model. This provides an aligned, traceable, and reusable temperature trajectory foundation for subsequent thermal margin stratification.

[0033] The controller uses the control cycle as the time reference. First, it performs time alignment and validity verification on the module temperature, coolant inlet temperature, coolant outlet temperature, ambient temperature, battery current, state of charge, vehicle power demand, and short-term driving status. Then, it writes the aligned signals into the same set of prediction input records. Subsequently, it calculates the short-term heat generation sequence based on the battery current and state of charge, and inputs the heat generation sequence and coolant temperature boundary into the lightweight thermal model. Through recursive calculation, it obtains the predicted temperature sequence and predicted hot spot temperature sequence of each module within the prediction window.

[0034] Therefore, aggregation and stratification are performed on a unified predicted temperature sequence to avoid amplifying sampling differences from different sources in the heat margin calculation.

[0035] During vehicle operation, module temperature, coolant inlet temperature, coolant outlet temperature, and battery current often originate from different sampling links, resulting in inconsistent sampling periods, timestamp jitter, and abnormal sampling patterns. Without prior alignment and verification, mismatches can occur in the predicted input, such as temperature originating from the previous cycle and current from the current cycle. This causes non-physical transitions in the lightweight thermal model at the recursion starting point, subsequently affecting the thermal margin stratification results. Therefore, at the beginning of each control cycle, the controller first establishes the observation record for that cycle, using the most recent valid value for time alignment, and performs validity verification using physical consistency constraints to ensure that the inputs entering the thermal model recursion are homogeneous and consistent.

[0036] The controller verifies the module temperature using the effective range of the module temperature, the coolant inlet and outlet temperatures using the consistency of the inlet-outlet temperature difference direction, and the battery current using the upper limit of the current change rate. If the verification fails, the controller replaces the corresponding observation with the effective observation from the previous control cycle and writes the replacement action into the diagnostic record so that subsequent steps can reference this record in the model reliability index. To reduce the step effect introduced by the hold-up strategy, the controller applies a first-order recursive filter to the replaced current and power demand signals, ensuring continuity of the predicted input between adjacent control cycles, thereby avoiding unnecessary excitation at the starting point of the thermal model recursion.

[0037] The controller performs time alignment of all observations with the control cycle as the boundary, completes the validity check with physical consistency constraints, and then writes the observations that pass the check into the prediction input record.

[0038] In practice, the predicted input records maintain consistency across both the temporal and physical dimensions, ensuring a stable starting point for the lightweight thermal model's recursion. This prevents thermal margin stratification from being affected by sampling mismatches. Diagnostic records synchronously save replacement actions, enabling subsequent steps to assess the applicability of the calibration window based on model reliability metrics.

[0039] Lightweight thermal model: indexed by cooling zones (Each partition corresponds to a branch valve and several modules), defining the partition characteristic temperature as the maximum value of the module temperature within that partition as the initial input value. Prediction Step The temperature recursion below can be performed using a first-order lumped model:

[0040]

[0041] Among them: zoned predicted temperature The above formula is derived recursively, with the initial value taken as the characteristic temperature of the current period's partition; the time step is... Control cycle, derived from the controller's timed task; partitioned thermal capacity. : Obtained through offline calibration, or by retrieving battery pack design parameters; regional heat generation prediction. Calculated from battery current and equivalent internal resistance (see next item); zoned equivalent heat transfer coefficient The initial values ​​for offline calibration can be obtained in the second step. The update indirectly corrects its local slope (i.e., the heat transfer trend near the current operating point); coolant inlet temperature The current inlet temperature can be maintained or extrapolated based on the short-term rate of change; both are feasible implementation methods.

[0042] The short-term heat generation intensity of a power battery is related to the battery current, state of charge, and internal resistance state, among which the internal resistance state drifts with changes in temperature and aging. If the heat generation is directly calculated using the static internal resistance constant, the predicted temperature series will show systematic deviations in different vehicle batches and different stages of use, thereby forcing the control strategy to introduce a larger safety margin.

[0043] Based on this, the controller introduces an internal resistance lookup table and interpolation process when constructing the heat generation sequence: using the state of charge and module temperature as lookup inputs, piecewise linear interpolation is used to obtain the equivalent internal resistance, which is then combined with the battery current to form the heat generation sequence, so that the heat generation calculation can change synchronously with the temperature and state of charge.

[0044] After obtaining the heat generation sequence, the controller uses a lightweight thermal model to recursively obtain the predicted temperature sequence. The lightweight thermal model uses the module temperature as the state variable, the heat generation sequence and the coolant inlet temperature as driving variables, and introduces the coolant outlet temperature as a heat transfer boundary check variable. When the predicted change direction of the coolant outlet temperature is inconsistent with the measured change direction, the controller restricts the value of the heat transfer coefficient to prevent the recursive result from conflicting with the energy conservation of the cooling loop. To ensure the recursive process can be implemented on the vehicle controller, the recursive calculation uses a fixed step size, consistent with the control cycle, and the prediction window consists of several consecutive control cycles. Each prediction step within the prediction window outputs the corresponding predicted module temperature, and the controller then obtains the predicted hotspot temperature sequence using a maximum value selection method, ensuring that the objects used for subsequent aggregation and stratification directly correspond to the hotspot constraints.

[0045] The controller uses the state of charge and module temperature to look up a table and interpolate to obtain the equivalent internal resistance (using a two-dimensional discrete grid table and bilinear interpolation to obtain the equivalent internal resistance). Then, it forms a heat generation sequence with the battery current. The heat generation sequence and the coolant temperature boundary are input into the lightweight thermal model for recursion, and the predicted temperature sequence and the predicted hot spot temperature sequence are output.

[0046] During use, the heat generation sequence is updated synchronously with changes in state of charge and temperature. The sources of deviation in the predicted temperature sequence are limited to traceable lookup and interpolation steps, thus making the thermal margin stratification more reflective of the actual thermal state of the vehicle. The thermal model recursion incorporates boundary checks on the coolant outlet temperature to ensure that the predicted trend is consistent with the energy transfer direction of the cooling circuit, avoiding non-physical temperature rises or drops.

[0047] Predicting hotspot temperature sequences involves multiple prediction steps. Directly taking the temperature of a certain prediction step as the basis for judgment can easily lead to frequent switching of layers due to single-step perturbations or changes in boundary conditions. On the other hand, directly taking the maximum value will amplify the single-step peak, resulting in a long-term small thermal margin, which in turn makes subsequent steps tend to be conservatively controlled.

[0048] Meanwhile, subsequent steps need to reflect both the heat margin of the risk of reaching the upper temperature limit and the trend strength of the hotspot approaching the upper temperature limit over time. Therefore, it is necessary to perform differentiable aggregation of the predicted hotspot temperature series, calculate the heat margin, and construct hotspot risk weights to ensure that the hierarchical criteria and control objectives have stability and transitivity.

[0049] Using the predicted hotspot temperature sequence as input, the system first constructs aggregated values ​​of predicted hotspot temperatures using logarithmic and exponential aggregation, ensuring that the aggregated values ​​are numerically close to the sequence peaks but not excessively dominated by spikes. Then, a heat margin is constructed based on the difference between the upper temperature limit and the aggregated values, and hotspot risk weights are constructed using this heat margin. Urgent and non-urgent zones are established according to the heat margin and temperature rise trend, and the system outputs upper temperature limit constraints, temperature difference constraints, minimum safe flow boundaries, and hotspot risk weights as unified inputs for subsequent steps.

[0050] The predicted hotspot temperature sequence is used to reflect the temperature peaks that may occur within the prediction window, but the peaks in the sequence may come from short-time current transitions or local spikes caused by thermal model boundary checks.

[0051] To achieve stable expression that avoids over-amplifying peaks while still indicating peak risk, a heat margin is constructed using logarithmic and exponential aggregation. This allows the sensitivity of the heat margin to sequence peaks to be adjusted by a single parameter while maintaining numerical continuity, facilitating stable use in subsequent steps for calibration window determination and risk weight calculation.

[0052]

[0053] Where: heat margin : indicates the first Temperature margin for each module within the prediction window , This indicates that the predicted peak temperature did not reach the upper limit. This indicates that the predicted peak temperature has reached or exceeded the upper temperature limit; upper temperature limit. : Indicates the highest permissible temperature threshold, with a value of Determined by battery safety strategy; aggregation coefficient : Represents the aggregation sensitivity parameter, with values ​​ranging from 0 to 10. , As the value increases, the aggregation result gets closer to the predicted sequence peak. When the value decreases, the aggregation result is closer to the average level of the predicted sequence; predicted temperature : indicates the first Each module is in the prediction step number The predicted temperature, taking values , derived recursively from the lightweight thermal model, is used to form a predicted hotspot temperature sequence;

[0054] Predicted steps : Represents the number of prediction steps within the prediction window, and is a positive integer; expressed as the aggregation coefficient. For predicted temperature Perform exponential aggregation and take the natural logarithm, then use the upper temperature limit. The heat margin is obtained by subtracting the polymerization result from the heat margin. .

[0055] Regional temperature rise trend :

[0056]

[0057] Urgent zone determination: If or If the condition is met, it is recorded as a pressing region; otherwise, it is recorded as a non-pressing region; where the heat margin threshold is... Temperature rise threshold Set the preset calibration constant (or look up the table segment by segment according to the ambient temperature), and specify that the source is offline calibration.

[0058] When in use, thermal margin The stratification criterion is determined by the temperature sequence of the entire prediction window, and does not depend on a single prediction step, thus avoiding frequent switching of stratification due to single-step perturbations. Aggregation coefficient. This makes the sensitivity of the polymerization results to peak values ​​controllable, so that the thermal margin can reflect peak risk without being dominated by peak error.

[0059] When heat margin is obtained The controller then converts the stratification results into urgent and non-urgent regions, as well as control objectives that can be directly referenced in subsequent steps. The urgent region is set as the priority option for meeting upper temperature limit constraints and temperature difference constraints, while the non-urgent region is set as the calibration time window for subsequent steps. The key to these two stratification criteria is that both thermal margin and other factors must be considered. The absolute level must be considered, as well as the directionality of the temperature rise trend within the prediction window, to avoid misjudging a non-urgent zone when the heat margin is not high but the temperature rises too quickly.

[0060] Therefore, the heat margin threshold and temperature rise trend threshold are used as hierarchical criteria, and hotspot risk weights are generated based on these criteria to prioritize the allocation of cooling resources in subsequent steps: constructing an effective heat margin. ,Again:

[0061]

[0062] Where: Hotspot risk weight : indicates the first Risk priority weights for each module , With heat margin Decrease and increase; risk coefficient : Represents the risk amplification factor, with a value of Used to adjust the risk weight of hot spots thermal margin The degree of sensitivity;

[0063] Marginal bias : Represents a bias to prevent the denominator from being zero, and its value is... Used to ensure hot spot risk weighting Numerical stability, and make It remains computable even when close to zero; heat margin Used to determine the degree to which risk weights are increased;

[0064] During the layer generation process, the controller first adjusts the heat margin based on the temperature rise trend. Perform a second judgment: when the heat margin When the temperature is below the thermal margin threshold and the temperature rise trend continues to increase, the module is marked as a critical zone; when the thermal margin... When the temperature exceeds the thermal margin threshold and the temperature rise trend becomes gradual, the module is marked as a non-critical zone. Subsequently, for the critical zone, immediate satisfaction requirements for the upper temperature limit constraint and temperature difference constraint are generated; for the non-critical zone, window markers allowing calibration actions are generated, and the minimum safe flow boundary is output simultaneously. The minimum safe flow boundary is jointly determined by the minimum circulation requirement of the cooling loop and the minimum heat exchange requirement of the cold plate. The controller expresses this as the minimum safe pump speed boundary and fixes the correspondence between this boundary and the coolant inlet temperature in the record, so that subsequent steps can directly call this boundary without recalculation.

[0065] As an example: When a vehicle enters a long, congested section of road under high ambient temperatures, the driver frequently starts and stops and intermittently accelerates. The battery management system continuously outputs module temperature and battery current, and the coolant inlet and outlet temperatures can be read on the instrument panel diagnostic interface. At the beginning of a control cycle, the controller aligns and verifies the observed measurements, then recursively generates a predicted hotspot temperature sequence within the prediction window and calculates the thermal margin for each module. Risk weighting of hot topics Modules with faster temperature rises after current sampling transitions are marked as being in the critical region, while modules with gradual temperature rises are marked as being in the non-critical region. This stratification result is written into the diagnostic log of the vehicle network, which also outputs the upper temperature limit, temperature difference limit, and minimum safe pump speed boundary for the current cycle. Engineers can see in the test log that the stratification markings are updated according to the operating conditions, and the critical region always corresponds to the thermal margin. Lower and hot topic risk weight Higher-level modules provide a direct basis for triggering calibration windows and allocation decisions in subsequent steps.

[0066] Furthermore, the controller adjusts the parameters based on the thermal margin. The thermal margin is calculated by dividing the urgent and non-urgent zones based on the temperature rise trend. Constructing Hotspot Risk Weights It outputs upper temperature limit constraints, temperature difference constraints, and minimum safe pump speed boundaries. During use, the stratification results, heat margin, and temperature rise direction are displayed synchronously. Switching between critical and non-critical zones is limited by the overall trend of the prediction window, preventing misjudgments based solely on instantaneous temperature. Hotspot risk weighting is also included. With heat margin It only performs a monotonic mapping, and the priority can be directly referenced when allocating cooling resources in subsequent steps.

[0067] Step 2: Using the heat margin stratification and control objectives given in Step 1, determine the permissible boundary of the calibration window. Within the calibration window, apply restricted perturbations to the pump speed and branch valve opening and analyze the temperature and electrical parameter responses. This will output the branch equivalent resistance coefficient, valve resistance deviation, heat transfer marginal slope, model confidence index, and anomaly indicators, thus providing a consistent and traceable set of parameter benchmarks for the branch flow distribution and pump operating point selection in Step 3.

[0068] Active self-calibration requires observable excitation, but the superposition of thermal and hydraulic inertia in a liquid-cooled system means that excessive excitation may introduce temperature fluctuations, while insufficient excitation may be overwhelmed by temperature noise and pump electrical parameter quantization errors. Simultaneously, the thermal margin output in the first step... Risk weighting of hot topics A unified characterization of safety margin and risk priority has been provided; therefore, [the following is a continuation of the previous sentence]. and The calibration is converted into a calibration permission boundary, and then the restricted disturbance is allocated to the corresponding branch in the non-urgent area. The pump speed is constrained by the minimum safe pump speed boundary, so that the boundary conditions of the calibration action and the marginal energy efficiency allocation in the third step are kept in the same terminology system.

[0069] The controller first reads the non-urgent zone markers output from the first step and filters out a set of candidate modules that can participate in the calibration; then, it maps the set of candidate modules to the corresponding branch valve openings. The candidate set is used to construct the calibration window permissive index by combining the minimum safe pump rate boundary and temperature difference constraint. It is used to determine whether the current control cycle enters the calibration window and the available amplitude of the current disturbance; after entering the calibration window, the controller arranges the disturbance in the order of first fixing the pump, then activating the valve, and then alternately activating the pump, so that the hydraulic parameters and heat transfer parameters exhibit separable characteristics in the response.

[0070] Whether calibration can be performed depends on whether the existing temperature margin is sufficient and whether the risk in this cycle is concentrated on a few urgent modules. If the calibration permission boundary is not clearly given, the controller may try to actuate the valve when the thermal margin is low, thereby tightening the temperature difference constraint and interfering with subsequent decisions.

[0071] Therefore, the controller will output the thermal margin in the first step. and hot topic risk weight Statistical calibration license index and use calibration license index Constraints can be placed on the set of valves participating in the calibration and on the upper limit of the applicable disturbance amplitude, thus limiting the calibration action to non-urgent areas and low-risk branches: Perform non-negativity truncation in the expression:

[0072]

[0073] Where: Calibration Permission Index : Indicates the permissible intensity of the calibration action, with a value of , A larger value indicates that the calibration action is more permissible; thermal margin : indicates the first The temperature margin of each module within the prediction window, taking the value... , used to characterize the temperature safety margin;

[0074] upper temperature limit : Indicates the highest permissible temperature threshold, with a value of Determined by battery safety strategy; hotspot risk weight : indicates the first Risk priority weight for each module, with a value of [value]. Used to characterize the degree of risk concentration; risk inhibition coefficient : Indicates the strength of suppression of risk weights, with values ​​ranging from 0 to 10. Used for regulation Sensitivity to risk accumulation;

[0075] In the selection of candidate branches, the controller uses the non-urgent area markers output in the first step as the first screening step to exclude branches corresponding to urgent areas; then, it uses hotspot risk weights... The sorting is the second filtering step, prioritizing the selection of... The smaller branch is designated as the excited branch, and the branch connected in parallel with it and sharing the manifold pressure drop is locked as the observation branch to distinguish hydraulic coupling. To prevent minimum cycle failure caused by disturbances, the controller uses the minimum safe pump speed boundary as the pump speed. The lower limit, and prohibit it within the calibration window. Below this boundary.

[0076] Furthermore, based on the calibration license index The threshold determination enters the calibration window, and is marked with non-urgent areas and hotspot risk weights. Complete the selection of the excited branch and the observed branch, and at the same time limit the pump speed with the minimum safe pump speed boundary. The lower bound.

[0077] In use, calibration actions are limited to stages with thermal margins and low-risk branches, thus preventing constrained disturbances from directly conflicting with upper temperature limits. The clear separation of the excited and observed branches allows subsequent parameter updates to distinguish between branch-specific changes and manifold-coupled changes. The minimum safe pump speed boundary affects the pump speed. The constraints ensure that the cooling circuit maintains basic circulation conditions.

[0078] After the calibration window is established, it is still necessary to solve the problem of how to excite the system so that the hydraulic and heat transfer characteristics can be distinguished in the observation. If the pump speed and the opening of multiple branch valves are changed at the same time, the system pressure drop and heat transfer boundary will change superimposed, and the parameter identification will have multiple solutions or strong correlations; if only one quantity is changed, the temperature change may be insufficient due to thermal inertia.

[0079] Based on this, a time-sharing arrangement is adopted: first, the pump speed is fixed. For the opening degree of a single branch valve Apply a restricted step change of rise-hold-recovery; then fix the valve opening and sequentially adjust the pump speed. Apply a restricted step with the same structure; and limit the number of valve actions, cumulative valve stroke and pump speed changes by limiting the action budget constraints, so that the calibration action can be performed for a long time under mechanical life and noise constraints.

[0080] The magnitude of the constrained disturbance is determined by the given allowable boundary, and the controller increments the valve opening by a certain amount. Select an integer multiple of the valve's minimum resolvable step size, and set the pump speed step size accordingly. The selected speed step is an integer multiple of the minimum speed step that the pump controller can stably maintain. The holding phase is counted with a fixed number of control cycles to ensure the comparability of observation segments. To enable analysis within the same observation window, the controller writes the start and end times of each rise-hold-recovery cycle into the observation segment marker and simultaneously records the module temperature, coolant inlet temperature, coolant outlet temperature, and pump electrical parameters within that segment, forming a continuous calibration observation record.

[0081] As an example: After a vehicle cruises at a constant speed on an urban ring road, it enters a long downhill coasting phase. The drive power demand decreases, the non-urgent zone marker output in the first step covers most modules, and the minimum thermal margin is achieved. This margin is significantly higher than the upper temperature limit constraint required. Based on this, the controller enters the calibration window, initially maintaining the water pump speed. Keeping the valve opening unchanged, adjust the opening degree of a branch valve located at the end of the battery pack. A minimum valve step is added and held for several control cycles before returning to the original valve position. During this holding period, the on-board diagnostics record shows a consistent change in the direction of coolant outlet temperature relative to inlet temperature, and a identifiable change in the slope of the temperature change of the corresponding module in that branch. The controller then maintains the branch valve opening. Keep the pump speed unchanged. Increase the speed step and hold it, then return to the original speed; during the water pump speed holding period, the water pump current changes consistently with the speed change, and the controller writes this change together with the coolant temperature response into the calibration observation record, and fixes the start and end times of each segment in the observation segment marker.

[0082] Furthermore, within the calibration window, the controller inputs the valve opening step size according to a time-division sequence of single-valve step-single-pump step. and water pump speed step By limiting the number of actions and cumulative strokes with action budget constraints, solidifying observation segment labels and calibrating observation records, and reducing the superposition relationship between hydraulic and heat transfer responses through time-sharing sequences, the updated parameters have a clear source of excitation. Action budget constraints ensure that valves and pumps maintain acceptable mechanical load and noise boundaries during long-term operation. Observation segment labels and calibrated observation records provide a continuous and traceable data basis for calculating residuals and updated parameters in the following sections.

[0083] After a restricted disturbance is applied, the system response includes changes in hydraulic network pressure drop, changes in coolant temperature boundary, and hysteresis effects caused by battery thermal inertia. If the observation segments are not decomposed according to the structural mechanism, parameter updates may misidentify thermal inertia as changes in branch resistance, or misidentify pump electrical parameter noise as valve characteristic drift.

[0084] Using the observation segment label as the boundary, residuals are constructed for the valve excitation segment and the pump excitation segment respectively, and the residuals are mapped to the update amount of the branch equivalent resistance coefficient, valve resistance deviation and heat transfer marginal slope. At the same time, the model confidence index and anomaly flag are used to distinguish between usable and unusable update results, so that the third step can automatically fall back to the conservative parameter set when the parameters are unreliable.

[0085] The calibration observation record is divided into valve excitation segment and pump excitation segment based on the observation segment label, and coolant boundary residual and module temperature slope residual are constructed in each segment. Secondly, the residual of valve excitation segment is mainly used to update valve resistance deviation and branch equivalent resistance coefficient, and the residual of pump excitation segment is mainly used to update the system equivalent pressure drop related quantities and back-calculate the consistency of branch equivalent resistance coefficient. Then, the consistency of the two types of update results is checked, the model credibility index is calculated and anomaly flags are generated, and the calibration is terminated and the update is frozen when necessary.

[0086] The controller limits the updated parameters to three categories: branch equivalent drag coefficient. Valve resistance deviation heat transfer marginal slope Among them, the equivalent resistance coefficient of the branch road Used to characterize the intensity of pressure drop variation with flow rate at a given coolant temperature in this branch; valve resistance deviation. Used to characterize valve opening. Shift in branch flow response; marginal slope of heat transfer This characterizes the direction and strength of the effect of increasing cooling flow near the current operating point on the slope of hotspot temperature change. Let the system pressure difference be... (Data from differential pressure sensor, or estimated from tables using pump speed and pump current); Total flow rate is (From the total flow sensor, or estimated from a table using pump speed and electrical parameters). For the first... Branch line definition of comprehensive hydraulic coefficient And adopt a simplified branch relationship:

[0087]

[0088] Then satisfy the parallel conservation: .

[0089] In the valve disturbance segment Known Known It is known that, therefore, updates can be performed on multiple perturbation segments using recursive least squares. Press again The output is the branch equivalent resistance coefficient. or valve resistance deviation At least one of the following.

[0090] In the valve disturbance segment, calculate the change in the slope of the hot spot temperature in the tight zone before and after the disturbance, and divide it by the branch to estimate the change in flow rate:

[0091]

[0092] Hotspot temperature slope It can be obtained by finite difference: ; From the above formula Pushed.

[0093] Within the valve actuation segment, the controller utilizes the difference in the slope of the module temperature change before and after the valve opening holding phase to obtain the marginal slope of the heat transfer. The local estimate is then used; subsequently, the change in coolant outlet temperature relative to inlet temperature, along with the change in pump electrical parameters, is used as an observed proxy for pressure drop / flow rate change, and this is used to correct for valve resistance deviation. This allows us to attribute the different responses resulting from the same valve opening change to valve characteristic drift or branch resistance drift.

[0094] To ensure on-vehicle feasibility, the controller uses recursive least squares as the update tool and projects the updated parameters onto the physically feasible region: branch equivalent drag coefficient. Valve resistance deviation is limited to a positive range. The heat transfer marginal slope is limited to the effective flow capacity that the valve can cover. The sign is defined as a physical direction that aligns with the direction of coolant inlet temperature change. If the vehicle is equipped with a differential pressure sensor, the differential pressure observation can directly replace the pump electrical parameters as a variable; if the vehicle is not equipped with a differential pressure sensor, the controller uses the water pump speed. The equivalent pressure difference was observed by constructing the pump current, and this construction method was fixed and written into the calibration observation record to ensure that subsequent diagnosis is traceable.

[0095] Specifically, valve excitation segments and pump excitation segments are analyzed separately according to the observation segment labels, and the equivalent resistance coefficients of the branches are updated using recursive least squares. Valve resistance deviation With the slope of heat transfer margin The updated results are then projected onto the physically feasible region and written into the parameter record. Valve resistance deviation. : Represents the resistance amplification deviation of the valve (including partial accessories) relative to its nominal state, with a value range of . , Indicates nominal resistance, The larger the value, the greater the resistance at the same valve opening degree.

[0096] When in use, the updated objects are limited to the hydraulic and heat exchange marginal parameters directly related to the third step, so that the parameter update and the subsequent marginal energy efficiency allocation form a one-to-one correspondence and reduce the implementation complexity; the combination of recursive least squares and physical feasible interval projection enables the parameter update to follow the drift while avoiding non-physical values; the equivalent implementation path of differential pressure sensor and pump electrical parameter proxy ensures that the same calibration process can still be executed under different hardware configurations.

[0097] After the parameters are updated, the controller needs to determine whether the update is reliable; otherwise, the third step may allocate flow and change the pump operating point under incorrect parameters. If the reliability judgment uses a simple average residual, it is easily misled by local spike errors; if variance or standard deviation indicators are used, statistical instability will occur in short segments.

[0098] Constructing a model credibility index To observe the residual sequence within the segment As input, a confidence expression sensitive to residual spikes but with continuous numerical values ​​is obtained by aggregating logarithms and exponentials, and then incorporated into the model confidence index. An exception flag is generated when the value is below the threshold. With the calibration termination flag, freeze this update and revert to the last trusted parameter record:

[0099]

[0100] Where: Model credibility index : Indicates the confidence level of the parameter update, and its value is... , The closer the value is to 1, the smaller the residual and the more reliable the update; sensitivity coefficient : Indicates the sensitivity to residual spikes, with values ​​ranging from 0 to 1. , It is more sensitive to larger residuals when the value increases;

[0101] residual : indicates the first The fragment residuals at each sampling point, taking values... The difference between the measured change in coolant outlet temperature within a segment and the model-predicted change, and the difference between the measured slope of module temperature within a segment and the model-predicted slope, can be synthesized with fixed weights to characterize segment consistency.

[0102]

[0103] in This represents the difference between the measured and model-predicted coolant outlet temperature. The difference between the measured slope of the hotspot temperature and the model prediction; weight. , This is a preset constant or normalized weight. , The source is either offline calibration or determined according to dimensional normalization. Sample size : Indicates the number of sampling points involved in the calculation within the segment, which is a positive integer and is determined by the observation segment label;

[0104] In the generation of exception flags, the controller will generate exception flags. This corresponds one-to-one with three types of phenomena: First, the model credibility index. If the value is below the threshold and persists for several control cycles, it is determined to be due to insufficient observation consistency; secondly, the valve opening in the valve excitation segment... The temperature has changed, but the coolant outlet temperature and water pump electrical parameters do not change accordingly, indicating a potential valve sticking or branch blockage; thirdly, the water pump speed in the pump excitation segment... The pump current change has changed, but the direction of the coolant temperature response is inconsistent, indicating an abnormal pump condition or significant air resistance. An abnormality indicator has appeared. Upon this, the controller immediately terminates the calibration window and freezes the current parameter update, while simultaneously flagging the anomaly. Write the diagnostic record so that the third step can be based on this to implement a degradation strategy and suppress further incentives.

[0105] With residuals Credibility Index of Sequence Computation Model Anomaly flags are generated by combining the response consistency between the valve excitation segment and the pump excitation segment. ,when Below the threshold or When valid, terminate the calibration and freeze the parameter records.

[0106] Model credibility index By combining logarithmic and exponential aggregation, both continuity and peak sensitivity are considered, making the credibility determination within short segments more stable and controllable. Anomaly flags. By establishing a correlation between insufficient reliability and operable field phenomena such as valve jamming, branch blockage, and air resistance, it is easier to directly adopt a conservative pump speed and valve position strategy in the third step. The termination and freeze mechanism ensures that the third step will not perform marginal energy efficiency allocation under low reliability parameters, thereby maintaining the consistency of the control chain.

[0107] Step 3: Under the premise that the upper limit of temperature and the temperature difference constraint are not exceeded, combine the heat margin. Risk weighting of hot topics Risk characterization and branch equivalent resistance coefficient Valve resistance deviation heat transfer marginal slope The circuit state is characterized to form the branch valve opening. The allocation results are used to determine the pump speed. The operating point is always subject to the minimum safe pump speed boundary. Valve action budget and model credibility index Abnormal signs Common constraints.

[0108] Branch valve opening Adjustments will simultaneously alter the flow distribution and manifold pressure drop distribution in parallel branches, thereby changing the slope of hotspot temperature changes and the pump load. If only the module temperature is used as the distribution basis, frequent valve position swings can easily occur near the temperature difference constraint boundary, and coolant will be delivered to branches that contribute less to hotspot temperature changes.

[0109] Since the heat transfer slope has already been output in the second step. Equivalent resistance coefficient of branch Furthermore, the first step has already output the hotspot risk weights. With heat margin Risk characterization, therefore , , and The results are aggregated into comparable discriminant quantities, and then transformed into an executable valve position adjustment sequence, ensuring a stable causal chain for valve position adjustment: first, identify branches that are more sensitive to changes in hot spot temperature and have lower hydraulic costs, and then gradually change the branch valve opening within the allowable range of valve action budget. .

[0110] The controller first checks the model confidence index. With abnormal flags When the abnormal flag Validity or Model Credibility Index When the value is below a preset threshold, the marginal decision is skipped and a conservative allocation is initiated; when the model credibility index... Acceptance criteria met and anomaly flags were displayed. When invalid, the controller treats the urgent zone marker as a hard constraint that prohibits reducing the valve opening, and the non-urgent zone marker as a permissible set that allows reducing the valve opening, thus forming an adjustable branch set.

[0111] Subsequently, a marginal energy efficiency criterion is constructed within the adjustable branch set, and the branch valve opening is adjusted accordingly. Perform monotonic updates: prioritize increasing valve opening for branches in the urgent zone with higher discrimination values. Subsequently, the valve opening was reduced for non-urgent branch circuits with lower discrimination values. At the same time, valve opening upper and lower limits, valve opening change rate and valve action budget constraints are applied to each valve opening update to avoid irreversible pressure drop transitions.

[0112] Marginal efficiency judgment does not simply write the rule of increasing flow rate as higher temperature, but rather combines the risk characterization in the first step with the marginal characterization in the second step into a comparable scalar, so that valve adjustment has a clear ranking basis.

[0113] To ensure that this criterion reflects hotspot risk, the marginal contribution of increased flow to hotspot temperature changes, and the hydraulic cost of branch channels, hotspot risk weights are used in this period. heat transfer marginal slope Branch equivalent resistance coefficient Valve resistance deviation Constructing the marginal energy efficiency discriminant The critical zone marker is embedded as a hard boundary for valve position adjustment, ensuring that the critical zone branch can only proceed towards increasing valve opening. Or maintain valve opening The direction is updated to avoid valve position swing-back on the branch with the highest risk:

[0114]

[0115] Where: Marginal energy efficiency criterion : indicates the first The priority of valve position adjustment for each branch in this cycle, and its value. , The larger the value, the more suitable the branch is for obtaining the increase in valve opening, which is used to determine the order of valve position updates.

[0116] Hotspot Risk Weighting : indicates the first Risk priority weights for each module ;

[0117] heat transfer marginal slope : Indicates the local influence of increasing the coolant delivery in the branch circuit on the slope of the hot spot temperature change, with a value of This is used to characterize whether increasing the valve position results in a perceptible change in hot spot temperature.

[0118] branch equivalent resistance coefficient : indicates the first The equivalent hydraulic resistance intensity of the branch is taken as a value. It is used to describe the hydraulic cost required to achieve the same transport change;

[0119] Valve resistance deviation : indicates the first The dimensionless deviation between the branch valve opening and the branch delivery response, with a value of , This indicates consistency with the nominal mapping. 0 indicates that the actual conveying capacity is reduced at the same valve opening degree. This indicates an increase in actual conveying capacity at the same valve opening degree, used to compensate for valve characteristic drift.

[0120] Regarding the embedding of urgency zone constraints, the controller does not treat the urgency zone marker merely as an explanation of the final allocation result, but instead writes the urgency zone marker into the valve position update rule: the branch valve opening of the urgency zone branch. It is forbidden to reduce the opening of branch valves in non-urgent areas. This allows for reduction within the valve's actuation budget. This rule makes the marginal energy efficiency criterion... The sorting results are consistent with the safety boundary to avoid conflicts where the valve position is reduced because the discriminant is low but the area is in a critical zone.

[0121] Calculate the marginal energy efficiency discriminant in this period The branch valve opening is constrained by marking the urgent zone. The update direction allows branch roads in the urgent zone to only be increased or maintained. Branch roads in non-urgent areas are allowed to be reduced in size. .

[0122] When used, the marginal energy efficiency discriminant is... By unifying risk characterization with marginal contribution and marginal cost within the same scalar, valve position adjustments have a comparable priority basis. Embedded constraints in the critical zone ensure that valve position reversal does not occur in the highest-risk branch, and the safety boundaries of upper temperature limit constraints and temperature difference constraints are maintained throughout the entire valve position update process. Dimensionless deviation. The introduction of this feature allows valve characteristic drift to be explicitly incorporated into the discriminant, preventing valve position adjustment from relying solely on nominal mapping while ignoring changes in operating conditions.

[0123] Marginal energy efficiency discriminant The sorting is provided, but the sorting itself cannot be directly converted into valve position commands because valves have upper and lower limits for opening, rate of change limits, and action budget constraints, as well as valve resistance deviation. This can cause different delivery variations on different branches when the same valve position is stepped. To ensure that valve position updates are both executable and interpretable, the controller employs a closed-loop method of monotonic stepping – reachability verification – re-stepping: first, the stepping amount is based on the valve opening. Increase the branch valve opening for the branch that is ranked higher. Then, using the same valve opening increments... Reduce the branch valve opening for branches that are later in the sequence and belong to the non-urgent zone. And after each update, the branch equivalent resistance coefficient is used. Valve resistance deviation Perform reachability checks to ensure that this update will not cause the manifold pressure drop to jump to the minimum safe pump speed boundary. It is difficult to maintain the cyclical state.

[0124] When updating the valve position, the controller sets the upper and lower limits of the operable valve opening for each branch, and the branch valve opening... The lower limit is used to ensure that the branch maintains basic coolant passage in non-urgent areas, and the branch valve opening degree The upper limit is used to prevent the valve from entering the end range where flow changes are not sensitive; at the same time, the controller limits the number of branches that can be modified and the total number of steps allowed in the current cycle based on the valve action budget, so that valve position updates do not result in high-frequency actions during the operating cycle. To ensure that the third step and the second step are physically causally closed, the controller marks the observed segment after the valve position update for use in the second step of the next control cycle, so that the parameter update of the next cycle can incorporate the response caused by the valve position change into a consistent calibration chain.

[0125] Furthermore, the marginal energy efficiency criterion is used to determine the quantity. The order of branch valve opening Perform monotonically incremental updates, and use the branch equivalent resistance coefficient after each update. Valve resistance deviation Perform reachability checks, and simultaneously apply upper and lower limits for valve opening, rate of change, and valve action budget constraints.

[0126] In use, valve position updates are performed in a step-by-step manner, ensuring that each valve position change has a clear source and a traceable observation window, facilitating subsequent diagnosis and determination of the correlation between valve position changes and temperature response. Accessibility verification checks hydraulic parameters... Deviation from valve characteristics A valve position update process is introduced to prevent the valve position sequence from becoming disconnected from the actual delivery response. Motion budget constraints limit the valve position update frequency to an acceptable range, thereby reducing valve wear and flow noise fluctuations.

[0127] water pump speed The choice of pump speed not only affects the total delivery capacity but also alters the efficiency range of the pump's operating point and influences the delivery distribution of parallel branches through manifold pressure drop. If a fixed pump speed or a fixed rule is used after valve position allocation, it's possible that the valve position has concentrated demand in the critical area, but the pump speed remains far above the minimum safe pump speed boundary. This situation causes the water pump to operate continuously in a low-efficiency or high-noise condition.

[0128] Since the model credibility index has already been output in the second step. With abnormal flags Parameter availability is used as a prerequisite for pump speed selection: when parameters are available, the pump speed... Based on the valve position allocation results, the parameters are checked step by step under the constraints of the high-efficiency operating range; when parameters are unavailable or abnormal indicators are displayed. When effective, the water pump speed It immediately switches to the conservative setting and freezes valve position updates to ensure that the upper temperature limit and temperature difference constraints are met first.

[0129] First, read the branch valve opening updated in the previous sub-step. As a fixed boundary, the minimum safe pump speed boundary As the water pump speed A hard lower limit was set, and the critical zone marker was used as a constraint to prevent a decrease in pump speed that would lead to a drop in the delivery speed of the critical zone branch; subsequently, the candidate pump speeds were determined based on the pump efficiency lookup table results. Verify each setting individually, prioritizing the lower setting that meets the threshold for high-efficiency operation; if a candidate setting does not meet the upper temperature limit or temperature difference constraint, slightly adjust the branch valve opening of the non-urgent zone branch. By changing the manifold pressure drop distribution, the candidate pump speed was re-evaluated. Then, if the problem persists, the pump speed is increased. Upgrade to a higher level. If an anomaly is detected at any stage of the verification process. Validity or Model Credibility Index If the water temperature drops below the threshold, immediately switch to a fallback mode: reduce the pump speed. Set to conservative position and open the branch valve. The system remains fixed at a conservative threshold until the second step of the next cycle yields an acceptable model credibility index. .

[0130] The high-efficiency operating range constraint does not require the introduction of additional hardware; instead, it uses pump efficiency as a verifiable boundary condition, embedding pump speed. The selection process. (Regarding pump efficiency) Considered to be determined by the water pump speed Lookup table values ​​indexed together with water pump electrical parameters: Water pump efficiency when the vehicle is equipped with a differential pressure sensor. The speed of the water pump Indexed in conjunction with differential pressure observation; water pump efficiency when the vehicle lacks a differential pressure sensor. The speed of the water pump This is indexed in conjunction with the pump current. The controller defines the high-efficiency operating range threshold as the efficiency threshold. And write it as a hard constraint or semi-hard constraint into the step-by-step verification process:

[0131]

[0132] Where: pump efficiency : Indicates the efficiency of the water pump at the current operating point, with a value of The efficiency can be obtained by looking up the table of water pump efficiency;

[0133] Efficiency threshold : Represents the minimum efficiency threshold for the high-efficiency work zone, with a value of [value missing]. The speed of the water pump is determined by the pump calibration data and is used to constrain the pump speed. Candidate tier selection;

[0134] In the step-by-step verification process, the controller operates at the minimum safe pump speed boundary. Starting with the lowest speed, candidate water pump speeds are selected one by one from the lowest to the highest speed. First, check the efficiency threshold for each candidate level. Then, the upper temperature limit constraint and temperature difference constraint are checked. To make the check process feasible, the controller uses piecewise linear interpolation to read the efficiency from the water pump efficiency table. The interpolation input only includes the pump speed. With a single observation index; when the interpolation result does not satisfy If the candidate gear is eliminated and the system moves to the next gear, it avoids using a high-speed, low-efficiency gear as the default selection. If all gears that meet the efficiency threshold fail to meet the upper temperature limit constraint, the controller allows the efficiency threshold to be treated as a semi-hard constraint. When the tight zone flag is active, the upper temperature limit constraint is prioritized, and a higher pump speed is selected. .

[0135] Furthermore, at the minimum safe pump speed boundary The above refers to the water pump speed. Each item is checked individually, and efficiency thresholds are used as a guide. Constraints on pump efficiency Within the acceptable range, piecewise linear interpolation is used to read the pump efficiency during the verification process. .

[0136] When in use, the efficiency threshold eliminates high-speed, low-efficiency gears from the candidate set in advance, thus reducing the pump speed. The selection no longer relies on fixed rules but has verifiable boundaries. Step-by-step verification is performed. Starting from this point, ensure the pump operating point is consistent with the minimum circulation requirement. Interpolation reading efficiency. Relying solely on the existing pump speed This allows for the reuse of the same constraint chain under different hardware configurations, in conjunction with electrical parameters or differential pressure observations.

[0137] When the model credibility index or abnormal flag When the indicated parameters are unavailable, continue to rely on the branch equivalent resistance coefficient. Valve resistance deviation With the slope of heat transfer margin Distributing resources will introduce uncontrollable errors into the pump and valve decision chain.

[0138] Therefore, downgrading as a safety net is considered an endogenous branch: on the one hand, if abnormal signs... Validity or Model Credibility Index If the opening is below the threshold, the branch valve opening for this cycle will be frozen. The update will adjust the branch valve opening of the emergency zone branch. Set the branch valve opening to the conservative opening range and adjust the branch valve opening of the non-urgent area branch. Set the pump within the range that maintains the opening of the passage; on the other hand, adjust the pump speed. Set to meet the minimum safe pump speed boundary And it should not be lower than the preset safety level, ensuring that the upper temperature limit and temperature difference constraints are met first. This fallback branch and the anomaly flag in the second step. One-to-one correspondence enables abnormal phenomena to directly trigger the visible pump and valve action sequence on site.

[0139] When there are no anomalies and the model credibility index Even when acceptable, the controller may still encounter a conflict during step-by-step verification where the upper temperature limit is difficult to meet but the efficiency threshold is too tight. In this case, the controller does not immediately increase the pump speed. Instead of upgrading to a higher level, a slight valve position adjustment is first implemented for the non-urgent zone branches: the marginal energy efficiency judgment value in the non-urgent zone is adjusted. Minimum branch valve opening Moving downwards, the marginal energy efficiency discrimination quantity in the critical area will be... Higher branch valve opening Step upwards to change the manifold pressure drop distribution, allowing limited delivery capacity to be more concentrated into the urgent branch lines.

[0140] Subsequently, the controller operates at the same candidate pump speed. The upper limit temperature constraint and temperature difference constraint are checked again; only when the valve position adjustment still cannot meet the constraints will the pump speed be increased to a higher level. The sequence of first adjusting the valve position and then increasing the pump speed ensures that the constraints of the pump's high-efficiency operating zone and the marginal energy efficiency distribution work in synergy: adjusting the valve position changes the hydraulic distribution, thereby reducing the number of pump speed increases required to meet the constraints, thus reducing the probability of the pump entering the low-efficiency or high-noise zone.

[0141] As an example: When a vehicle climbs a hill continuously in high summer temperatures, the initial output of the urgency zone markers is concentrated in the central module of the battery pack, indicating a hotspot risk weight. The corresponding branches are prioritized; the second step, which involved calibration and outputting the equivalent drag coefficients of the branches during the previous cruise phase, was completed earlier. Valve resistance deviation With the slope of heat transfer margin And the model credibility index Acceptable, Abnormal Flags Invalid. After the controller enters this step, it first determines the marginal energy efficiency. Increase the opening of the branch valves in the corresponding branches of the middle module. At the same time, reduce the opening degree of branch valves in non-urgent areas. It is also subject to valve action budget constraints; the driver can observe the water pump speed on the vehicle status page. The valve position changes switch between several discrete levels rather than continuously increasing, and these changes are concentrated in a few branches rather than all branches operating simultaneously. The controller then operates at the minimum safe pump speed boundary. The efficiency of the water pump is checked step by step above. With efficiency threshold If a certain setting meets the efficiency threshold but the upper temperature limit constraint is still close to the boundary, the controller will first adjust the opening degree of the branch valve in the non-urgent zone. And recalibrate, and increase the water pump speed if necessary. Ultimately, this manifests as the branch valves in the emergency zone remaining at a high opening, while the branch valves in the non-emergency zone remain at the opening of the maintenance path, and the pump speed... Maintain the operation within a verifiable range. If valve sticking occurs during this process and an abnormality flag is triggered... The controller freezes the valve position update and adjusts the water pump speed within a control cycle. Switching to the conservative setting resulted in the valve position no longer changing and the pump speed remaining constant. Maintain stability.

[0142] Model credibility index With abnormal flags As a branch condition, the branch valve opening is frozen under abnormal conditions. Update and increase the water pump speed Set to the conservative setting; under acceptable conditions, first adjust the pressure drop distribution by adjusting the valve position in the non-urgent zone, and then verify the candidate pump speed. Only increase the pump speed if the constraints still cannot be met. Gear shift.

[0143] Downgrading and providing a safety net will be an abnormal sign. Directly corresponding to the visible pump and valve actions, the upper temperature limit constraint and temperature difference constraint of the abnormal process have a guaranteed path. Valve position return prioritizes changing the pressure drop distribution, and the water pump speed... Upgrading is no longer the only option; reducing the time the pump spends in low-efficiency or high-noise settings is crucial. The implementation examples consistently show that most valve position changes occur in a few branches with limited operational budgets, facilitating on-site maintenance and diagnostics.

[0144] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0145] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0146] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0147] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0148] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A flow rate regulation method for a power battery liquid cooling system based on operating condition prediction, characterized in that: include, Collect module temperature, coolant inlet and outlet temperature, current and state of charge, and vehicle operating condition information. Based on the lightweight thermal model, calculate and predict the temperature trajectory and obtain the predicted temperature rise trend. Calculate the thermal margin and divide the urgent zone and non-urgent zone. Output the upper limit temperature constraint, temperature difference constraint and hot spot risk weight. When the heat margin meets the preset threshold and the predicted temperature rise trend is gradual, a limited disturbance is applied to the pump speed and branch valve opening in the non-critical zone, and the battery temperature and coolant temperature difference are monitored. The equivalent resistance parameters and heat transfer margin slope are updated based on the disturbance-response. Based on the hot spot risk weight, temperature upper limit constraint, temperature difference constraint, and combined with the updated equivalent resistance parameter and heat transfer marginal slope, the branch valve opening is first allocated and then the pump speed is determined. The hot spot temperature change corresponding to the unit pump work is used as the allocation metric and the pump high-efficiency working area constraint is satisfied. Collect ambient temperature and water pump speed electrical parameters, perform time alignment and validity verification on module temperature and coolant inlet and outlet temperatures, write the aligned signal into the prediction input record, replace abnormal signals with the most recent valid value and write them into the diagnostic record, and further output the minimum safe pump speed boundary. The heat margin is obtained by performing logarithmic and exponential aggregation operations on the predicted temperature trajectory within the prediction window, and negative heat margins are truncated. The predicted temperature rise trend is obtained by the temperature difference between adjacent prediction steps, and the urgent zone and the non-urgent zone are jointly determined by the heat margin and the predicted temperature rise trend. The lightweight thermal model uses discrete recursion with control cycle as the step size to generate the predicted temperature trajectory. The heat generation is calculated by current and equivalent internal resistance. The equivalent internal resistance is obtained by bilinear interpolation of a two-dimensional table of state of charge and module temperature index. The consistency of the recursion boundary is checked by the coolant outlet temperature. The calibration window is triggered by the thermal margin meeting the preset threshold and the predicted temperature rise trend meeting the condition of being gradual. After triggering, the branch with the smaller hot spot risk weight is selected from the non-urgent area as the excited branch, and the branch sharing the manifold with it is set as the observation branch, and the pump speed is not lower than the minimum safe pump speed boundary.

2. The flow rate regulation method for a power battery liquid cooling system according to claim 1, characterized in that: The restricted disturbance is applied in a time-sharing sequence. First, a step change, hold, and recovery are performed on the opening of a single branch valve while keeping the pump speed constant. Then, a step change, hold, and recovery are performed on the pump speed while keeping the branch valve opening constant. The step change amplitude is an integer multiple of the actuator's minimum resolvable step and is constrained by the valve action budget.

3. The flow rate regulation method for a power battery liquid cooling system according to claim 2, characterized in that: The disturbance-response consists of electrical parameters such as module temperature, coolant inlet and outlet temperature, coolant temperature difference, and water pump speed within the observation segment to form a calibration observation record. The residual is synthesized by the residual of coolant outlet temperature change and hot spot temperature change slope residual according to preset weights, and the model credibility index is generated accordingly.

4. The flow rate regulation method for a power battery liquid cooling system according to claim 3, characterized in that: The updated parameters also include the valve opening-flow relationship deviation. The equivalent resistance parameter and the valve opening-flow relationship deviation are limited to a preset physical feasible range through projection mapping. Anomaly flags are calculated based on the model credibility index and the perturbation-response consistency. When the anomaly flag is in a valid state, the calibration window is terminated and the update is frozen.

5. The flow rate regulation method for a power battery liquid cooling system according to claim 4, characterized in that: The allocation of branch valve opening is based on the hot spot temperature change corresponding to the unit pump work to construct the allocation priority, and the accessibility is checked by combining the hot spot risk weight, heat exchange marginal slope, equivalent resistance parameter and valve opening-flow relationship deviation; the branch valve opening corresponding to the urgent zone is prohibited from being reduced, and the branch valve opening is adjusted in a monotonic step manner and meets the valve action budget.

6. The flow rate regulation method for a power battery liquid cooling system according to claim 5, characterized in that: The high-efficiency operating range constraint of the water pump is achieved by looking up the water pump efficiency table. The water pump efficiency is obtained by interpolation based on the water pump speed and system pressure difference. The system pressure difference is collected by the differential pressure sensor and estimated by looking up the water pump speed electrical parameter information when the differential pressure sensor is unavailable. The water pump speed is checked step by step from the minimum safe pump speed boundary. If the check fails, the branch valve opening is adjusted first and then the water pump speed is adjusted.

Images