Keyboard performance intelligent detection method, terminal equipment and storage medium

Abstract

The invention relates to the technical field of electronic equipment performance detection, in particular to an intelligent keyboard performance detection method, terminal equipment and a storage medium. In the intelligent keyboard performance detection method, a space-time binding data set of pressure, stroke and contacts is constructed through three-dimensional signal coupling acquisition, and through double-channel processing of a dynamic behavior rule base, the pressure, stroke and contact information is obtained; wherein the channel I identifies and marks manual operation behaviors based on a legal operation feature model, and eliminates fault misjudgment caused by operation noise; the channel II strips marked data through an unmatched signal screening device, extracts pure hardware features to generate an independent physical signal flow, and solves data pollution of operation noise to aging modeling; and finally, hardware abnormal data is mapped to a substrate conduction path through a physical conduction rule base, and a contact fatigue contribution output region fatigue weight model is fused, so that precise diagnosis of hardware faults and reliable prediction of the service life of the substrate are realized, and double defects of misjudgment and aging prediction distortion are radically eliminated.

Description

Technical Field

[0001] This invention relates to the field of electronic device performance testing technology, specifically to an intelligent keyboard performance testing method, terminal device, and storage medium. Background Technology

[0002] Current keyboard performance testing solutions face significant limitations when dealing with complex user operation scenarios. For example, traditional methods rely on single-point physical signal threshold judgment mechanisms (such as triggering a fault alarm when the key conduction time exceeds the limit), making it difficult to effectively distinguish between user-initiated operations (such as long-pressing of combination keys in gaming scenarios) and actual hardware malfunctions (such as contact point jamming). Especially in high-frequency operation areas of combination keys, human operation signals and hardware fault signals overlap in time and frequency domain characteristics, resulting in a persistently high false alarm rate. Meanwhile, continuous operational noise contaminates the original physical signals of the hardware, causing a large number of false wear characteristics to be mixed into the collected data (such as transient pressure peaks generated by high-intensity impacts being misjudged as substrate fatigue). This leads to aging prediction models based on historical data deviating from the actual hardware degradation trajectory. This detection blind spot coupled with operational interference and hardware status creates a dual technical dilemma: a simultaneous decline in the accuracy of fault diagnosis and the reliability of life prediction. This not only increases unnecessary maintenance costs but may also mask the true risk of hardware degradation. Summary of the Invention

[0003] The purpose of this invention is to provide a keyboard performance intelligent detection method, terminal device, and storage medium to solve the problems mentioned in the background art. Specific technical problems include how to eliminate operational interference, address the technical defect of misjudging hardware freezes due to human operation (such as long-pressing of key combinations), and achieve accurate hardware fault diagnosis; and how to construct a reliable aging model based on pure hardware signals to solve the problem of substrate fatigue prediction distortion caused by operational noise pollution.

[0004] To achieve the above objectives, one objective of this invention is a keyboard performance intelligent detection method, comprising the following steps: S1. A spatial hierarchical compression strategy is implemented through a pressure sensor array. Based on the stress distribution of the key positions, the core high-frequency region, the mid-frequency region, and the edge low-frequency region are divided. The highest sampling frequency is used in the high-frequency region (the intersection of the direction key and the function key) to capture the transient pressure change gradient. The mid-frequency region uses a medium sampling frequency to track the average pressure trend. The lowest sampling frequency is used in the low-frequency region to record the steady-state distribution. This strategy enables the transient pressure coupling effect of the key combination operation to be captured with high precision, providing a high-resolution data foundation for subsequent behavior recognition.

[0005] The design simultaneously acquires the dynamic curve of key travel using a Hall effect sensor, accurately determines the critical position of the travel by setting a magnetic saturation monitoring point under the keycap, and records the depth increment during the pressing phase and the reset delay during the rebound phase. When an abnormally flat area of ​​the rebound curve is detected, a displacement-time function matrix characterizing mechanical wear is generated. This design enhances the ability to detect travel sticking points and provides reliable trajectory data for hardware wear analysis.

[0006] The conductive film contact adopts active impedance detection technology, injecting a square wave current pulse at the moment of contact conduction. Based on the pulse decay waveform, the asymmetric oscillation envelope is identified to determine the impedance change of the metal oxide layer, and the spatiotemporal coordinates of the change point and the impedance increment are recorded simultaneously. This technology breaks through the limitations of traditional on / off detection, which expands the contact on / off time sequence into a metal fatigue time sequence dataset, providing key parameters for early warning of contact oxidation failure.

[0007] The three signals are aligned by a unified clock source to generate a three-dimensional coupled signal set, forming a spatiotemporally bound data body of pressure, stroke, and impedance, providing structured input for dual-channel processing.

[0008] S2. Input the three-dimensional coupled signal set into the dynamic behavior rule library for time-series dual-channel processing. Channel 1 calls the legal operation feature model (including game scene pressure distribution patterns and tremor waveform library) to perform matching and parsing operations on the three-dimensional coupled signal set. When the pressure peak at a specific coordinate continuously exceeds the limit and the slope fluctuation period of the travel curve conforms to the tremor feature, the legal operation behavior judgment engine is triggered. The pressure distribution pattern of adjacent coordinates is associated through the spatiotemporal window diffusion algorithm, and the feature label sequence with spatiotemporal coordinate labels is output. This step accurately distinguishes between human operation (such as long press of game combination keys) and hardware failure, eliminating the misjudgment problem caused by operation noise in traditional solutions.

[0009] In Channel 2, based on the feature-labeled sequence, a spatiotemporal mapping table is constructed using an unmatched signal filter to label and permanently remove data segments covered by legitimate operations. Three types of clean hardware features—pressure gradient anomalies, travel jamming points, and impedance abrupt changes—are extracted from the unlabeled intervals to generate independent physical signal flows. The unmatched signal filter simultaneously runs a self-optimization mechanism, generating a spatial contact fatigue heatmap by extracting contact response delay data from the same topological coordinates. It records the operation type and duration percentage of the removed segments and updates the confidence level of the operation feature model based on the statistical features of the removed data. This process completely removes operation noise, allowing the hardware physical state to be extracted without contamination, while continuously improving the accuracy of behavior recognition through learning.

[0010] S3. The feature-labeled sequence and independent physical signal streams are input into the physical conduction rule base. The sequence-signal mapper parses the topological coordinate network of high-frequency operating events to locate the substrate stress input source nodes. The conduction path matcher maps hardware anomaly data from the independent physical signal streams to conduction path nodes based on parameters such as the substrate's elastic modulus distribution. Simultaneously, it integrates the conduction efficiency impact characteristics of the spatial contact fatigue thermal map, converting metal fatigue parameters into node efficiency attenuation coefficients. The stress-deformation coupling algorithm integrates path efficiency attenuation, mechanical deformation accumulation, and contact fatigue contribution to output a regional fatigue weight model with risk level labels. This step constructs a full-dimensional aging prediction model based on clean hardware data, achieving accurate positioning of substrate fatigue areas and graded early warning of failure risks.

[0011] The second objective of this invention is to provide a terminal device, including a processor and a memory, wherein the memory stores a computer program, and when the program is executed, it implements an intelligent keyboard performance detection method. Specifically, it solves the real-time processing requirements that traditional solutions cannot meet by using a collaborative design of an embedded processor and a multi-level memory. The processor integrates a dedicated signal acquisition and control unit, directly connecting to the pressure sensor array, Hall sensor group, and conductive thin-film contact interface to achieve hardware-level synchronous acquisition and clock alignment of three signals. The memory employs a hierarchical architecture of high-speed cache and non-volatile storage, with the dynamic behavior rule base and physical conduction rule base residing in the high-speed cache to ensure latency-free scheduling during dual-channel processing. The physical modeling accelerator implements the stress-deformation coupling algorithm using hardware logic circuits, reducing computation time. This device, through direct sensor interface connection, pre-loading of the rule base in hardware, and algorithm accelerator integration, forms a complete hardware closed loop from signal acquisition to model output, overcoming the real-time bottleneck of software solutions in complex signal processing and meeting the millisecond-level response requirements of online keyboard monitoring.

[0012] The third objective of this invention is to provide a storage medium for storing computer programs, which, when executed, implement an intelligent keyboard performance detection method. Specifically, this involves designing a multi-layered instruction structure to achieve efficient deployment and cross-platform reuse of the technical solution, wherein: The first instruction layer encapsulates the sensor-driven control logic, directly calling the underlying hardware resources of the pressure sensor, Hall sensor, and conductive thin-film contacts through the operating system kernel interface to generate a time-aligned three-dimensional coupled signal set. The second instruction layer compiles the dynamic behavior rule library into a dynamically loadable binary module, supporting time-sharing preemptive scheduling of dual-channel processing tasks to ensure the parallel generation of feature marker sequences and independent physical signal streams. The third instruction layer, with the physical conduction rule library at its core, uses a GPU-accelerated instruction set to achieve parallel computation of sequence-signal mapping, conduction path matching, and stress-deformation coupling algorithms. This medium, through instruction pipeline optimization and hardware abstraction layer design, allows the same program body to be adapted to different processor architectures, solving the portability and compatibility issues of detection algorithms on heterogeneous hardware platforms.

[0013] Compared with the prior art, the beneficial effects of the present invention are: This invention achieves precise separation of operational behavior noise and hardware physical state through a dual-channel processing mechanism based on a dynamic behavior rule base. Channel 1 identifies legitimate operational behaviors based on a scenario-based feature model and generates a feature marker sequence, eliminating misjudgments caused by operations such as long press of combination keys. Channel 2 uses an unmatched signal filter to strip away the marked data fragments, extracts pure hardware features, and generates an independent physical signal stream, solving the problem of data pollution caused by operational noise in aging modeling.

[0014] In addition, by combining spatial hierarchical compression strategy to improve the accuracy of high-frequency signal capture, active impedance detection technology to expand the dimensions of contact fatigue monitoring, and rule base self-optimization mechanism to continuously improve the accuracy of behavior recognition, a full-link optimization from signal acquisition to model output is formed.

[0015] Ultimately, the physical conduction rule base is used to achieve the fusion modeling of multi-dimensional hardware features such as pressure mutation, stroke stagnation, and impedance anomaly, and outputs a regional fatigue weight model with risk level, which improves the accuracy of fault diagnosis and the reliability of life prediction, provides core technical support for proactive keyboard maintenance, significantly reduces unnecessary maintenance costs and extends equipment life. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall method of the present invention; Figure 2 This is a performance test comparison chart of the present invention. Detailed Implementation

[0017] The technical solutions in 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.

[0018] Next, please refer to Figure 1 One of the objectives of this embodiment is to provide a keyboard performance intelligent detection method, which includes the following steps: S1. Pressure distribution cloud map acquisition is performed through a pressure sensor array covering the keyboard substrate. Its innovation lies in the use of a spatial hierarchical compression strategy, specifically including: First, based on the mechanical stress distribution characteristics of the keys, the substrate is divided into three monitoring areas: the core high-frequency area (covering the intersection of the directional keys and function keys), which bears high-frequency impact loads, uses the highest sampling frequency to capture microsecond-level pressure gradient changes in real time; the mid-frequency area (center of the main key area) uses a medium sampling frequency for regular keystrokes to track the trend of average pressure changes; and the edge low-frequency area (such as above the numeric keys) uses the lowest sampling frequency to record the steady-state pressure distribution; among them: The core high-frequency zone is located at the physical intersection of the arrow keys and function keys. This area is subjected to high-frequency impact loads, so it is configured with the highest sampling frequency (specifically, to capture microsecond-level pressure change gradients in real time). The mid-frequency zone covers the mechanical center of the main key area and is configured with a medium sampling frequency for regular keystrokes (specifically, to track the trend of average pressure changes). The edge low-frequency zone is selected above the mechanical structure of the numeric keypad and is configured with the lowest sampling frequency (specifically, to record the steady-state pressure distribution). This spatial hierarchical compression strategy is determined based on the actual load frequency characteristics of the regions. The core high-frequency zone needs to capture transient pressure coupling effects due to frequent combination key operations, the mid-frequency zone uses balanced sampling for standard keystroke operations, and the edge low-frequency zone uses sparse sampling due to low usage.

[0019] Furthermore, this spatial hierarchical compression strategy optimizes data effectiveness through partitioned differential sampling. Specifically, dense sampling points in the high-frequency zone can capture the transient pressure coupling effect of key combination operations (such as the pressure peak transmission path when pressing Ctrl+C simultaneously), while sparse sampling in the mid- and low-frequency zones reduces redundant data, providing a spatially optimized compressed pressure cloud map data volume for the dynamic behavior rule base.

[0020] A magnetic induction array based on Hall sensors acquires the dynamic curve of key travel in real time. Each sensor is embedded under the keycap and generates a linear Hall voltage change with key movement, specifically including: The key travel depth increment is recorded during the keycap pressing phase, and the reset response delay is tracked during the rebound phase. In particular, a magnetic saturation monitoring point is set at the end of the key travel to accurately determine the critical position of the travel corresponding to the moment the contact is turned on. The key travel dynamic curve records the acceleration, deceleration characteristics and travel inflection point coordinates of the key movement in a time series manner, forming a displacement-time function matrix that can reflect the degree of mechanical wear of the keycap. When an abnormally flat area (such as a stuck point) is detected in the key travel rebound curve, this data will be used as the key time trajectory in the subsequent three-dimensional coupling signal to characterize physical wear, that is, the key travel dynamic curve.

[0021] The conductive film contacts utilize active impedance detection technology to generate a contact on / off timing sequence, specifically including: At the instant the contact is physically turned on, a square wave current pulse with a pulse width ≤1ms is injected into the thin-film circuit, and the pulse decay waveform is captured by a high-speed ADC (analog-to-digital converter). The instantaneous impedance value is calculated according to the formula Z=(V_peak·τ) / ∫i(t)dt (where Z is the contact impedance, ∫i(t)dt is the integral of current over time, V_peak is the peak voltage, τ is the pulse width, and t is the timestamp). When an asymmetric oscillating envelope is detected in the decay waveform, a sudden change in the metal oxide layer impedance is determined. The timestamp of the sudden change point, the contact coordinates, and the impedance increment are recorded simultaneously. This technology breaks through the limitations of traditional on / off detection, so that the contact on / off sequence not only includes the switching state, but also extends to a metal fatigue time series dataset that includes impedance decay characteristics.

[0022] The pressure distribution cloud map, key travel dynamic curve, and contact on / off timing sequence are aligned with a unified clock source and then subjected to spatiotemporal dynamic binding, specifically including: First, establish the association rule between the pressure distribution cloud map and the contact on / off timing sequence. When a pressure peak occurs at a certain coordinate in the pressure distribution cloud map within a specific time window, if the contact on / off sequence at the same coordinate has a conduction event within the time tolerance range, then bind the contact conduction time with the pressure peak time. Simultaneously associate the conduction point depth and the slope of the key travel dynamic curve, and finally generate a three-dimensional coupled signal set. The data structure is a sequence of quadruples containing coordinate, pressure peak attribute, travel depth attribute, impedance attribute, and key travel dynamic curve slope attribute.

[0023] S2. Input the three-dimensional coupled signal set into the dynamic behavior rule base and perform time-series dual-channel processing. Channel one inputs the four-tuple sequence from the three-dimensional coupled signal set into the legal operation feature model for dynamic matching and parsing, specifically including: First, the scenario-based feature matrix pre-built in the behavior rule library is invoked, including a high-frequency game operation mode library (such as the pressure in the direction key area continuously exceeding the preset pressure threshold, and asymmetric pressure gradient distribution features) and a user behavior tremor waveform library (the slope of the key travel dynamic curve fluctuates periodically within a specific range). Based on the legal operation feature model, the four-tuple sequence in the three-dimensional coupled signal set is traversed. When a specific coordinate is detected to meet the condition that the pressure peak continuously exceeds the limit and the fluctuation period of the key travel dynamic curve slope matches the preset tremor feature, the legal operation behavior judgment engine is triggered to perform a matching and parsing operation. This engine uses a spatiotemporal window diffusion algorithm to associate the pressure distribution morphology features of adjacent coordinate intervals, and finally outputs a structured feature label sequence—the data structure is a set of tuples of {start timestamp, end timestamp, coordinate set, behavior type code}. Among them, the behavior type code strictly inherits the operation category defined by the hierarchical region (such as the game combination key long press operation code).

[0024] Channel 2 uses the feature marker sequence as the spatiotemporal coordinate basis for the operation, and performs precise data separation through an unmatched signal filter, specifically including: Using the feature-labeled sequence as the spatiotemporal coordinate basis for the operation, precise data separation is performed through an unmatched signal filter: Establish a spatiotemporal mapping table for filtering, and mark the spatiotemporal blocks covered by legitimate operations within the three-dimensional coupled signal set based on the timestamp intervals and coordinate sets in the feature marker sequence; The dynamic signal truncation operation physically isolates all quadruple data within the marked spatiotemporal block. If the pressure peak occurs during the marked time period and the coordinates belong to the marked coordinate set, the data is permanently removed. Three types of hardware physical feature data were extracted from the unlabeled spatiotemporal interval: pressure abrupt gradient anomaly (pressure value changes abruptly and is not associated with operation marker), travel stagnation point (the slope of the dynamic curve of button travel is lower than the activity threshold and continues to time out), and impedance abrupt value (the amount of impedance change of metal oxide layer exceeds the material critical value). The final result is a pure hardware state dataset that does not contain human interference, which is generated as an independent physical signal stream. It retains the original data structure but achieves physical purification.

[0025] In addition, during the signal separation process of Channel 2, the unmatched signal filter extracts the response delay data of continuous contact points with the same topological coordinates, associates the pressure distribution attenuation gradient to generate a spatial contact point fatigue heat map; simultaneously records the feature label type (such as game operation code) and duration ratio of the eliminated segments, and updates the confidence weight of the legal operation feature model of the dynamic behavior rule base based on the statistical distribution characteristics of pressure / stroke in the eliminated segments.

[0026] S3. Input the feature-labeled sequence and independent physical signal flow into the physical conduction rule base to perform collaborative modeling, specifically including: The sequence-signal mapper first parses the topological coordinate network of operation events in the feature-marked sequence and extracts the spatiotemporal coordinate distribution thermal features of the high-frequency operation area; the conduction path matcher maps the three types of hardware physical feature data in the independent physical signal flow to the corresponding coordinate nodes of the conduction path network based on the material stress conduction characteristic parameters (including elastic modulus distribution and interlayer bonding strength gradient) preset on the keyboard substrate. The spatial contact fatigue thermal map generated by S2 is input into the conduction path matcher as the spatial distribution benchmark for impedance mutation values. Based on the previously generated spatial contact fatigue thermal map data, the conduction efficiency influence characteristics and contact state weighting factors at each coordinate position are extracted. The physical parameters characterizing metal fatigue in the thermal map are transformed into efficiency attenuation coefficients of the substrate conduction path nodes. At the same time, the spatial distribution enhancement correction of impedance mutation characteristics is performed according to the contact state weighting factors. Based on the stress conduction characteristics of the substrate material, the conduction path matcher maps the three types of physical characteristic data after fatigue characteristic correction to the conduction path coordinate nodes. Finally, the stress-deformation coupling algorithm integrates the path efficiency attenuation, mechanical deformation accumulation, and contact fatigue contribution to generate a partitioned substrate region fatigue weight model that integrates the contact fatigue mechanism. The stress-deformation coupling algorithm performs multi-physical quantity collaborative calculation, specifically including: First, a node physical state matrix of the substrate conduction path network is established. The dimensions of this matrix include the path efficiency attenuation output by the conduction path matcher (characterizing the loss of current conduction efficiency caused by the fatigue of the conductive film metal), the mechanical deformation accumulation (reflecting the amount of substrate plastic deformation caused by keyway jamming), and the contact fatigue contribution (quantifying the weighted impact of the contact oxide layer impedance change on the overall fatigue). Spatial convolution operation is performed on the node matrix using weighted convolution kernels. The weights of the convolution kernels are dynamically configured based on the stress transmission characteristics of the substrate material: the elastic modulus distribution determines the propagation coefficient of the mechanical deformation accumulation, the interlayer bonding strength gradient controls the diffusion range of path efficiency decay, and the metal oxidation sensitivity adjusts the weight ratio of the contact fatigue contribution. The convolution output is normalized to generate the fatigue accumulation index of each node. Finally, based on the spatial topology of the nodes, the partition weights are allocated to form a regional fatigue weight model that includes the substrate partition coordinates, fatigue index weight values, and material failure risk level labels.

[0027] Please see Figure 2 This invention demonstrates its technical advantages through a dual-channel processing mechanism. In the field of keyboard performance detection, the collaborative operation of a dynamic behavior rule base and independent physical signal streams achieves a dual technological breakthrough, wherein: As shown by the solid green line in the figure, the fault false positive rate of this invention remains consistently below 2.1% (at 100,000 samples on the horizontal axis), a 94.5% reduction compared to the fixed threshold detection method (solid red line). This is attributed to the effective filtering of operational noise such as long presses of combination keys by the dynamic rule base. Through real-time analysis of key behavior patterns, it exhibits stable noise immunity characteristics at a sample size of 40,000 (marked as the point where the "dual-channel processing mechanism is effective").

[0028] Regarding the aging prediction accuracy (yellow solid line), the present invention's scheme steadily improved to a plateau of 93.5% as the sample size increased, which is 40.8 percentage points higher than the fixed threshold detection method (blue dashed line). Especially at 70,000 samples, when the fixed threshold detection method experienced a precipitous drop due to operational noise pollution (marked as "operational noise pollution data"), the pure hardware model constructed by the present invention through physical signal flow maintained a stable upward trend.

[0029] The implementation results validated the core innovative value of the invention, achieving physical isolation between operational behavior and hardware wear and tear. When traditional solutions suffer from a surge in misjudgment rate (38.2%) and a collapse in accuracy (52.7%) due to data contamination, the dual-channel architecture of the invention keeps both key indicators within the optimal range, providing an anti-interference, high-precision industrial-grade solution for keyboard reliability monitoring.

[0030] The second objective of this embodiment is to provide a terminal device, including a processor and a memory, wherein the memory stores a computer program, and when the program is executed, it implements a keyboard performance intelligent detection method, specifically including: The terminal device includes an embedded processor and a multi-level memory architecture. The processor integrates three core modules: a signal acquisition and control unit, a rule base processing engine, and a physical modeling accelerator. The memory architecture adopts a hierarchical storage strategy. The cache stores the legal operation feature models of the dynamic behavior rule base and the substrate characteristic parameters of the physical conduction rule base. The non-volatile memory chip stores the signal processing algorithm instruction set. When the device starts keyboard performance detection, the processor synchronously triggers the data acquisition circuits of the pressure sensor array, Hall sensor group, and conductive thin film contacts through the sensor interface. The raw signal stream is input to the signal acquisition and control unit in real time to perform clock alignment and three-dimensional coupling signal set construction. The rule base processing engine calls the feature model in the cache to perform time-series dual-channel processing on the three-dimensional signal set, generating feature label sequences and independent physical signal streams. The physical modeling accelerator performs sequence-signal mapping and path matching calculations based on the conduction rule base, and finally generates a regional fatigue weight model in the processor memory and outputs it to the display interface or network transmission module.

[0031] The third objective of this embodiment is to provide a storage medium for storing a computer program, which, when executed, implements an intelligent keyboard performance detection method, specifically including: The computer-readable storage medium is a solid-state storage chip or an optical storage carrier. The computer program stored therein contains a multi-layer instruction structure: the first instruction layer implements signal acquisition and control logic, which generates a clock-aligned three-dimensional coupled signal set by calling the underlying operating system driver to control the synchronous sampling of pressure sensors, Hall sensors, and conductive thin-film contacts; the second instruction layer loads a dynamic behavior rule base to perform dual-channel processing, where the first instruction set calls a legal operation feature model to perform pattern matching on the signal set and outputs a feature label sequence, and the second instruction set constructs an unmatched signal filter based on the label sequence to generate an independent physical signal stream; the third instruction layer activates the physical conduction rule base processing engine, which parses the event topology network through sequence-signal mapping instructions, maps abnormal data based on substrate conduction parameters, and outputs a region fatigue weight model through stress-deformation coupling instructions; during program execution, the pipeline transmission of instructions at each layer is realized through a memory buffer. When a keyboard performance analysis instruction is detected, the processor loads the instruction set from the storage medium layer by layer into the cache, forming a complete program execution chain from signal coupling to fatigue modeling.

[0032] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent keyboard performance detection, characterized in that, The methods and steps include the following: S1. The pressure distribution cloud map of the button is collected by the pressure sensor array, the dynamic curve of the button travel is obtained by the Hall sensor, the contact on and off timing sequence is recorded by the conductive film contact, and the three signals are aligned by a unified clock source to form a three-dimensional coupled signal set. S2. Input the three-dimensional coupled signal set into the dynamic behavior rule base and perform time-series dual-channel processing, wherein channel one performs matching parsing on the three-dimensional coupled signal set based on the legal operation feature model and outputs a feature label sequence. In channel two, the feature marker sequence is used as the filtering basis. Unmarked signal segments are separated from the three-dimensional coupled signal by the unmatched signal filter to generate independent physical signal streams. S3. Input the feature marker sequence and the independent physical signal flow into the physical conduction rule base. The sequence-signal mapper parses the event topology coordinates. The conduction path matcher maps the abnormal data to the corresponding coordinates according to the substrate conduction characteristic parameters. Finally, the stress-deformation coupling algorithm outputs the regional fatigue weight model.

2. The intelligent keyboard performance detection method according to claim 1, characterized in that, In step S1, a spatial hierarchical compression strategy is used when collecting the key pressure distribution cloud map, specifically including: Based on the mechanical stress distribution characteristics of the bonds, the substrate is divided into a core high-frequency region, a mid-frequency region, and an edge low-frequency region. The highest sampling frequency is used to capture the pressure abrupt gradient in the core high-frequency region, a medium sampling frequency is used to track the pressure mean change trend in the mid-frequency region, and the lowest sampling frequency is used to record the steady-state pressure distribution in the peripheral low-frequency region.

3. The intelligent keyboard performance detection method according to claim 1, characterized in that, The step S1 of obtaining the key travel dynamic curve includes: The Hall sensor embedded under the keycap records the incremental travel depth during the pressing phase and the reset response delay during the rebound phase; a magnetic saturation monitoring point is set at the end of the key travel to determine the critical travel position; a displacement-time function matrix containing acceleration, deceleration characteristics and travel inflection point coordinates is generated; when an abnormally flat area is detected in the rebound curve, it is used as a dynamic curve of key travel characterizing physical wear.

4. The intelligent keyboard performance detection method according to claim 1, characterized in that, The step S1, which generates the contact on / off timing sequence, employs active impedance detection technology, specifically including: A square wave current pulse is injected at the moment the contact is physically turned on, and the instantaneous impedance value is calculated by capturing the pulse decay waveform. When an asymmetric oscillating envelope is detected in the decay waveform, a sudden change in the impedance of the metal oxide layer is determined. The timestamp of the impedance change point, the contact coordinates, and the impedance increment are recorded simultaneously.

5. The intelligent keyboard performance detection method according to claim 1, characterized in that, The execution process of Channel 1 in step S2 specifically includes: The system calls a pre-built legal operation feature model, which includes a scenario-based feature matrix and a user behavior tremor waveform library. When a specific coordinate in the three-dimensional coupled signal set satisfies the condition that the pressure peak continuously exceeds the limit and the slope fluctuation period of the key travel dynamic curve conforms to the tremor feature, a matching parsing operation is performed to trigger the legal operation behavior judgment engine. The engine then associates the pressure distribution morphology features of adjacent coordinate intervals through a spatiotemporal window diffusion algorithm. The output includes a feature tag sequence that includes a start timestamp, an end timestamp, a coordinate set, and a behavior type encoding.

6. The intelligent keyboard performance detection method according to claim 1, characterized in that, The execution process of channel two in step S2 includes: A spatiotemporal mapping table is established based on the feature label sequence to mark the spatiotemporal blocks covered by legitimate operations in the three-dimensional coupled signal set; The data within the marked spatiotemporal blocks are physically isolated by the unmatched signal filter, permanently removing data whose pressure peaks occur during the marked time period and belong to the marked coordinate set. Three types of hardware physical feature data—pressure abrupt gradient anomaly, travel stagnation point, and impedance abrupt value—are extracted from unlabeled spatiotemporal intervals to generate an independent physical signal stream that retains the original data structure but achieves physical purification.

7. The intelligent keyboard performance detection method according to claim 6, characterized in that, The unmatched signal filter executes synchronously during operation: Extract response delay data of continuous contact points with the same topological coordinates, associate them with pressure distribution attenuation gradient to generate spatial contact fatigue heat map; record the feature label type and duration ratio of the eliminated segments; update the confidence weight of the legal operation feature model based on the statistical distribution characteristics of pressure and stroke in the eliminated segments.

8. The intelligent keyboard performance detection method according to claim 1, characterized in that, The operation of the transmission path matcher in step S3 includes: Map the hardware physical characteristic data in the independent physical signal flow to the corresponding coordinate nodes of the conduction path network; By utilizing the influence characteristics of conduction efficiency at each coordinate position in the spatial contact fatigue thermogram and the contact state weighting factor, the physical parameters of metal fatigue are transformed into the efficiency attenuation coefficient of the substrate conduction path node. Spatial distribution enhancement correction of impedance mutation characteristics is performed based on contact state weighting factors; the path efficiency attenuation, mechanical deformation accumulation and contact fatigue contribution are integrated through stress-deformation coupling algorithm.

9. A terminal device, comprising a processor and a memory, wherein the memory stores a computer program that, when executed, implements the keyboard performance intelligent detection method as described in any one of claims 1-8.

10. A storage medium storing a computer program that, when executed, implements the intelligent keyboard performance detection method as described in any one of claims 1-8.

Images