Multi-modal intelligent household appliance intelligent level quantification evaluation method

Abstract

The embodiment of the application provides a kind of quantification evaluation method for multi-modal intelligent household appliance intelligent level, comprising: constructing household appliance intelligent performance evaluation model, simulating first environmental change process, starting household appliance equipment to identify first environmental change process to make household appliance equipment obtain environmental data;Collect the first data processing cycle process in household appliance equipment, evaluate the first data processing cycle process, generate the first evaluation report;Make household appliance equipment execute environmental regulation action, identify second environmental change process, evaluate the environmental regulation capability of household appliance equipment according to second environmental change process, generate the second evaluation report;According to the method proposed in the embodiment of the application, a unified, quantitative method can be established to benchmark test the intelligence of household appliances, and fair comparison between different household appliance types and models is met.

Description

Technical Field

[0001] This application relates to, but is not limited to, the field of home appliance performance evaluation technology, and in particular to a quantitative evaluation method for the intelligence level of multimodal smart home appliances. Background Technology

[0002] The proliferation of smart home devices has revolutionized home life, embedding advanced sensing, data processing, and connectivity into everyday appliances. Today, devices such as thermostats, refrigerators, washing machines, and lighting systems behave in accordance with user preferences, optimize energy consumption, and interact seamlessly within an interconnected ecosystem. Despite these advancements, a significant gap remains in standardized, objective methods for assessing and comparing the intelligence levels of these diverse appliances, which currently rely on subjective user experiences. Current assessment methods are often qualitative, depending on subjective user experiences or manufacturer self-claims, failing to capture the varying information processing capabilities of modern devices. This lack of a unified assessment framework hinders informed consumer choices and impedes manufacturers' efforts to systematically enhance the intelligence of their appliances. Summary of the Invention

[0003] The following is an overview of the subject matter described in detail herein. This overview is not intended to limit the scope of the claims.

[0004] This application provides a quantitative evaluation method for the intelligence level of multimodal smart home appliances, which can establish a unified and quantitative method to benchmark the intelligence of home appliances and meet the requirements for fair comparison between different types and models of home appliances.

[0005] To achieve the above objectives, a first aspect of this application proposes a quantitative evaluation method for the intelligence level of multimodal smart home appliances, comprising: constructing a smart home appliance performance evaluation model, simulating a first environmental change process, initiating the home appliance to identify the first environmental change process so that the home appliance can acquire environmental data; collecting a first data processing loop process within the home appliance, evaluating the first data processing loop process, and generating a first evaluation report; causing the home appliance to perform environmental control actions, identifying a second environmental change process, evaluating the home appliance's environmental control capability based on the second environmental change process, and generating a second evaluation report.

[0006] In some embodiments, the step of collecting a first data processing loop within the home appliance, evaluating the first data processing loop, and generating a first evaluation report includes: reading the data types acquired by the home appliance during the first data processing loop and their corresponding environmental data; constructing a data type matrix based on the acquired data types; evaluating the home appliance's ability to identify environmental states based on the data type matrix; and solving for the unknown environmental state before the home appliance perceives the environmental data to obtain environmental entropy; constructing an environmental information matrix based on the environmental data; evaluating the home appliance's ability to perceive environmental states again based on the environmental information matrix to obtain a first conditional entropy; judging the home appliance's ability to perceive the unknown environmental state using the environmental information matrix based on the environmental entropy and the first conditional entropy; determining the home appliance's first environmental perception ability; and generating a first evaluation report based on the first environmental perception ability.

[0007] In some embodiments, the step of collecting a first data processing loop within the home appliance, evaluating the first data processing loop, and generating a first evaluation report further includes: reading the encoding structure of the home appliance during the first data processing loop, and the original features obtained from the environmental state perceived by the sensors; determining the encoding distortion degree of the home appliance for environmental state information based on the difference between the original features and the encoded data under the encoding structure; evaluating the authenticity retention of the encoding structure of the home appliance during the first data processing loop based on the original features and the encoding structure, and determining the encoding entropy difference; and analyzing the utilization efficiency of the home appliance for environmental state information based on the encoding distortion degree and the encoding entropy difference, and generating a first evaluation report.

[0008] In some embodiments, the step of causing the home appliance to perform environmental control actions, identifying a second environmental change process, evaluating the home appliance's environmental control capabilities based on the second environmental change process, and generating a second evaluation report includes: reading a set of action instructions inferred by the home appliance based on its internal knowledge after receiving environmental data; evaluating the home appliance's ability to control the environment under the influence of the internal knowledge based on the set of action instructions; determining a second conditional entropy; mapping the environmental data to the set of action instructions; causing the home appliance to select and execute environmental control actions based on the set of action instructions; obtaining the expected environmental state predicted by the home appliance based on the internal knowledge; determining an environmental state difference based on a set target environmental state and the expected environmental state; and evaluating the capabilities of the internal knowledge and the performance of the home appliance based on the second conditional entropy and the environmental state difference, thereby generating a second evaluation report.

[0009] In some embodiments, the step of causing the home appliance to perform environmental control actions, identifying a second environmental change process, evaluating the home appliance's environmental control capabilities based on the second environmental change process, and generating a second evaluation report further includes: the home appliance acquiring initial environmental information and controlled environmental information after completing the environmental control actions; the home appliance updating its internal knowledge based on the initial environmental information and the controlled environmental information; evaluating the degree of updating of the internal knowledge; determining the knowledge gain of the internal knowledge; and generating a second evaluation report.

[0010] In some embodiments, evaluating the update degree of the internal knowledge, determining the knowledge gain of the internal knowledge, and generating a second evaluation report includes: gradually updating the internal knowledge over a period of time, sequentially determining the changes in discrete parameters of the internal knowledge before and after each update node, determining whether the internal knowledge has converged based on the changes in discrete parameters, and determining the learning optimization efficiency of the internal knowledge; determining the knowledge gain between the internal knowledge based on the internal knowledge before and after adjacent update nodes; and generating a second evaluation report based on the learning optimization efficiency and the knowledge gain.

[0011] In some embodiments, the method further includes: simulating a third environmental change process, controlling a group of home appliances to adjust to the third environmental change process, wherein the group of home appliances includes multiple home appliances that are interconnected, and the group of home appliances is set in the same test space; collecting a second data processing loop within the group of home appliances, detecting the interconnection status within the group of home appliances; and generating a third evaluation report based on the interconnection status.

[0012] In some embodiments, generating a third evaluation report based on the interconnection status includes: collecting the data interaction efficiency of the home appliance group during the interconnection process; acquiring the first environmental information utilization efficiency of the home appliances during the first environmental change process; collecting the second environmental information utilization efficiency of the home appliances during the third environmental change process by transmitting environmental information between the home appliances; determining a first group effect based on the first and second environmental information utilization efficiencies; detecting the first environmental adjustment effect of the home appliances on the environment after the second environmental change process; detecting the second environmental adjustment effect of the home appliance group on the environment after the third environmental change process; determining a second group effect based on the first and second adjustment effects; and generating a third evaluation report based on the data interaction efficiency, the first group effect, and the second group effect.

[0013] To achieve the above objectives, a second aspect of this application proposes an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the quantitative evaluation method for the intelligence level of multimodal smart home appliances described in the first aspect.

[0014] To achieve the above objectives, a third aspect of this application proposes a storage medium storing a computer program that, when executed by a processor, implements the quantitative evaluation method for the intelligence level of multimodal smart home appliances described in the first aspect.

[0015] The embodiments of this application include at least the following beneficial effects: the first evaluation report generated based on the first data processing cycle of the home appliance provides a quantitative standard for the intelligent performance of the home appliance, while meeting the user's deep needs for the autonomous management function of the device; in addition to the data processing cycle within the home appliance, by controlling the home appliance to perform environmental regulation actions and analyzing the second environmental change process, the dynamic adaptability of the device in actual application is verified, ensuring that it can continuously optimize the regulation strategy based on feedback. Combined with the second evaluation report, it can help to locate problems and iteratively improve. By analyzing the correlation between regulation results and user needs, the home appliance can gradually realize the prediction and satisfaction of undefined needs, promoting the evolution of the intelligent system towards a higher level of proactive service.

[0016] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description

[0017] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0018] Figure 1 This is a schematic diagram of an optional process for a quantitative evaluation method of the intelligence level of multimodal smart home appliances provided in an embodiment of this application. Figure 2 A schematic diagram of an optional process for evaluating environmental sensing provided in an embodiment of this application; Figure 3 An optional flowchart illustrating the evaluation process of the internal coding representation provided in an embodiment of this application; Figure 4 An optional flowchart illustrating the evaluation reasoning and decision-making process provided in an embodiment of this application; Figure 5This is a schematic diagram of an optional process for evaluating adaptive update learning provided in an embodiment of this application; Figure 6 A schematic diagram of an optional process for evaluating interconnection provided in an embodiment of this application; Figure 7 This is a schematic diagram of an optional structure of the smart performance evaluation device for home appliances provided in an embodiment of this application; Figure 8 This is a schematic diagram of an optional hardware structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0020] In the description of this application, "several" means one or more, "multiple" means two or more, "greater than", "less than", "exceeding" etc. are understood to exclude the number itself, and "above", "below", "within" etc. are understood to include the number itself.

[0021] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, or the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0022] In related technologies, current evaluation methods are typically qualitative, relying on subjective user experience or manufacturer self-claims, failing to capture the diverse information processing capabilities of modern devices. The lack of a unified evaluation framework makes it difficult for consumers to make informed choices and hinders manufacturers from systematically improving the intelligence of home appliances.

[0023] Based on this, the embodiments of this application provide a quantitative evaluation method for the intelligence level of multimodal smart home appliances, which can establish a unified and quantitative method to benchmark the intelligence of home appliances and meet the requirements for fair comparison between different types and models of home appliances.

[0024] The quantitative evaluation method for the intelligence level of multimodal smart home appliances provided in this application is specifically illustrated through the following embodiments. First, the quantitative evaluation method for the intelligence level of multimodal smart home appliances in this application embodiment is described.

[0025] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0026] The embodiments of this application will be further described below with reference to the accompanying drawings.

[0027] like Figure 1 As shown, Figure 1 This is an optional flowchart illustrating a quantitative evaluation method for the intelligence level of multimodal smart home appliances provided in this application embodiment. This method can be executed by a server, a terminal, or a server in conjunction with a terminal. The method includes, but is not limited to, the following steps S110 to S130: Step S110: Construct a smart performance evaluation model for home appliances, simulate the first environmental change process, and activate the home appliances to identify the first environmental change process so that the home appliances can acquire environmental data. Step S120: Collect the first data processing loop process within the home appliance, evaluate the first data processing loop process, and generate a first evaluation report; Step S130: The home appliance performs environmental control actions, identifies the second environmental change process, evaluates the home appliance's environmental control capabilities based on the second environmental change process, and generates a second assessment report.

[0028] Information theory provides a solid theoretical foundation for addressing this challenge by offering mathematical tools for quantifying uncertainty, information flow, and adaptability. This application's embodiments view the functions of intelligent systems in home appliances (data acquisition, representation, reasoning, learning, and decision-making) as a unified process of reducing uncertainty and optimizing results. This defines intelligence as a high-level form of information processing. To quantify the "intelligence" of home appliances into metrics that can be computed using information theory, this application's embodiments propose an information theory framework that deconstructs the intelligence of home appliances into quantifiable evaluation dimensions. Then, it applies basic information theory methods to numerically evaluate each dimension, ultimately establishing a unified, quantitative method for benchmarking the intelligence of home appliances and ensuring fair comparisons between different types and models of home appliances.

[0029] The intelligent performance of home appliances can be broken down into the ability to acquire, transform, and utilize information to act autonomously, efficiently, and adaptively. At the heart of this definition is the ability of home appliances to reduce uncertainty about their environment or operating conditions, making decisions that align with user preferences, environmental constraints, and performance goals. For users, this manifests in three levels of needs: First, intelligent technologies that respond to environmental and user data enable devices to autonomously manage their functions, minimizing the need for human intervention; second, personalized services are provided based on individual user differences: device functions are customized to suit each user's unique preferences and circumstances, ensuring a customized and adaptive user experience; and finally, big data and advanced analytics are used to predict and meet user needs that may not be explicitly recognized or expressed.

[0030] Based on the quantitative evaluation method for the intelligence level of multimodal smart home appliances proposed in this application, this application constructs a unified evaluation model to provide a benchmark environment for subsequent testing, ensuring the consistency and comparability of data collection, and providing a foundation for fair evaluation of cross-model and cross-brand devices. By simulating the first environmental change process and activating the device identification function, the responsiveness of home appliances under different environmental conditions can be comprehensively tested. The evaluation of the data processing loop can accurately identify the capabilities of home appliance algorithms in data representation, reasoning, and learning, integrating the data processing of home appliances into quantifiable and repeatable indicators. Based on the first data processing of home appliances... The first evaluation report generated during the cyclical process provides a quantitative standard for the intelligent performance of home appliances, while also meeting users' deeper needs for autonomous management functions. In addition to the data processing within the home appliances, by controlling the appliances to perform environmental control actions and analyzing the second environmental change process, the dynamic adaptability of the devices in practical applications is verified, ensuring that they can continuously optimize control strategies based on feedback. Combined with the second evaluation report, this helps in identifying problems and iteratively improving them. By analyzing the correlation between control results and user needs, home appliances can gradually predict and meet undefined needs, driving the intelligent system towards a higher level of "proactive service." The method proposed in this application constructs a closed-loop evaluation system that deeply integrates the information theory framework with user needs, ensuring that the intelligent performance of home appliances in each stage of data collection, processing, and control is quantifiable, and achieving fair comparison across device types.

[0031] Information theory employs three computational tools: entropy, mutual information, and KL divergence. These tools measure data uncertainty, the degree of dependency between variables, and the difference in probability distributions, respectively. Entropy quantifies the disorder of data distributions, guiding feature selection and diversity control in generative models. Mutual information assesses the strength of the association between two variables, used for feature selection and multimodal alignment. KL divergence compares the degree of deviation between two distributions. Throughout the entire process from data preprocessing to model optimization, entropy, mutual information, and KL divergence help the system cope with uncertainty, capture key associations, and correct biases, ensuring the robustness of the algorithm in complex tasks.

[0032] From an information theory perspective, smarter home appliances should include the following intelligent directions: accurately acquiring relevant environmental data and user signals to reduce uncertainty; effectively encoding, storing, and retrieving important information, preserving key insights for future use; drawing meaningful inferences, making predictions, and selecting actions to optimize outcomes, using internal models to guide decision-making even under uncertain conditions; continuously learning and adapting, improving their internal performance and strategies as conditions change or new information emerges; and integrating and utilizing external resources, such as cloud-based knowledge bases and communication with other devices, to enhance their decision-making capabilities.

[0033] To grasp the multifaceted nature of smart home appliances and in accordance with the aforementioned information theory-based approach to intelligence, this application defines several core dimensions that constitute the complete information lifecycle of smart home appliances, from raw data collection to action selection and continuous adaptation, quantifying the efficiency of home appliances in reducing uncertainty and utilizing available information.

[0034] First, embodiments of this application propose an evaluation of the information acquisition and physical perception dimensions, focusing on the device's ability to perceive its environment. High intelligence requires accurate and reliable sensors to capture the state of the world and user input with minimal uncertainty. Information-theoretic metrics (e.g., the mutual information between sensor output and ground reality) indicate how effectively the device's perception capabilities reduce environmental ambiguity. Figure 2 As shown, Figure 1 Step S120 includes, but is not limited to, steps S210 to S230: Step S210: Read the data type and corresponding environmental data acquired by the home appliance in the first data processing loop, construct a data type matrix based on the acquired data type, evaluate the home appliance's ability to identify the environmental state based on the data type matrix, and solve the unknown environmental state before the home appliance senses the environmental data to obtain the environmental entropy. Step S220: Construct an environmental information matrix based on environmental data, and re-evaluate the ability of home appliances to sense environmental status based on the environmental information matrix to obtain the first conditional entropy; Step S230: Based on environmental entropy and first conditional entropy, evaluate the ability of home appliances to perceive unknown environmental states using the environmental information matrix, determine the first environmental perception capability of the home appliances, and generate a first evaluation report based on the first environmental perception capability.

[0035] In the field of smart home appliances, information acquisition and physical sensing refer to the ability of home appliances to capture and process environmental data in real time. This data typically includes various physical parameters such as temperature, humidity, pressure, and light intensity, which can be sensed using various sensors. However, due to the differences in the design, cost, and intended functions of home appliances, they are usually equipped with different types of sensors, each with varying sensitivity, accuracy, and coverage.

[0036] The core challenge of the proposed approach lies in quantifying the extent to which these sensors reduce environmental uncertainty and how much relevant environmental information they capture, especially when sensors differ in sensor type, data diversity, and fusion capabilities. This concerns not only accuracy but also sensor diversity and how this diversity can be effectively integrated to contribute to a comprehensive understanding of the environment.

[0037] The method proposed in this dimension conducts in-depth analysis of the types of data collected by home appliances and the environmental data they respond to, and evaluates the diversity of sensors. The types of sensors in home appliances may include temperature, humidity, pressure, light, motion, and sound sensors, etc. The multimodal data combination of various different and cross-domain sensors can help home appliances to have a unified understanding of environmental conditions. In addition, home appliances can also connect to external devices and acquire cloud data to enhance the perception capabilities of home appliances, thereby gaining a deeper understanding of environmental conditions and user behavior.

[0038] First, in this embodiment of the application, a matrix E is defined to represent the environmental information collected by the home appliance. Each element in the data type matrix E represents a different environmental parameter and condition: .

[0039] This represents measurable environmental data, where i is an integer with a value in the range [1, n], and n represents the number of perceived environmental states.

[0040] By modeling environmental conditions and sensor data in matrix form, complex multidimensional environmental information and sensor outputs are organized into a structured format for further analysis and processing. Taking an intelligent air conditioning system as an example, it relies on multiple sensors to monitor different environmental conditions and make corresponding adjustments. Air conditioning sensors may include: temperature sensors (monitoring room temperature), humidity sensors (monitoring room humidity), light sensors (monitoring indoor light intensity), and motion sensors (detecting whether someone is in the room). These sensors collect data in real time, and the state of the environment at each moment can be viewed as a vector. This data can be organized and represented in matrix form for further analysis.

[0041] Suppose that the environmental parameters that home appliances focus on are temperature, humidity, light intensity, and motion. We organize these environmental parameters into a matrix according to time and sensor type. For example, the home appliance collects environmental data at two time points, with each time point containing these four parameters. The matrix can be represented as follows: .

[0042] Let represent the temperature at time i, where i is an integer and its value ranges from [1, 2]. Let i represent the humidity at time i, where i is an integer and its value ranges from [1, 2]. Let represent the light intensity at time i, where i is an integer and its value ranges from [1,2]. This represents the motion detection value at time i, where i is an integer and its value ranges from [1,2]. The motion detection value is used to indicate whether there is a person present. For example, when the motion detection value is 0, it means that no one is present, and when the motion detection value is 1, it means that someone is present.

[0043] In this example, each row of the matrix represents the environmental state at a specific point in time, while each column represents the output of different sensors at that point in time.

[0044] At the same time, home appliances acquire environmental data through sensors. Here, i is an integer, and its value ranges from [1, m]. An environmental information matrix is ​​constructed using the readings from m sensors. .

[0045] By modeling the environmental state that home appliances can obtain based on sensors, after the home appliances acquire environmental data, the environmental information matrix S is compared with the macroscopic actual environmental conditions to determine the amount of reduction in uncertainty of the home appliances themselves in the first environmental change process, thereby quantifying the home appliances' ability to capture environmental state.

[0046] Suppose that the home appliance contains multiple different sensors, such as a temperature sensor. Humidity sensor Light sensor and motion sensors Each sensor has different outputs at different time points. Assuming there are two time points, organize this data into a matrix according to sensor and time: .

[0047] in, This represents the output of the i-th sensor at the j-th time point. At this time, i and j are both integers, with i ranging from [1,4] and j ranging from [1,2].

[0048] In practical applications, smart home appliances can use matrix modeling methods to process and analyze data collected from different sensors. For example, in air conditioning systems, matrix models can help analyze the following: By integrating data from various sensors (temperature, humidity, light, etc.), the air conditioning system can comprehensively assess the comfort level of a room and make decisions accordingly, such as adjusting the temperature or turning the air conditioner on or off. Through analysis of matrix data, the system can identify different environmental patterns, such as changes in environmental conditions during the day and night, or when the room is occupied and unoccupied. Based on these patterns, the system can optimize its efficiency, such as implementing an energy-saving mode. Using historical data matrices, the system can predict future environmental changes and make adjustments in advance. For example, if the system detects fluctuations in temperature and humidity over the past few hours, it can pre-adjust to ensure the room reaches optimal comfort when the user enters.

[0049] This application's embodiments define environmental entropy. To represent the uncertainty of the environmental state, that is, the amount of information needed to describe the environmental state, the following is the formula for environmental entropy: .

[0050] in the formula Represents the i-th environmental information The probability distribution, where i is an integer, and the entropy of the environmental parameter E, i.e., the "uncertainty" or "information" of the system. Environmental parameter This refers to a specific environmental variable sensed by home appliances through sensors. These environmental parameters may include temperature, humidity, light intensity, motion detection, etc., all of which are values ​​collected in real time by sensors. These environmental parameters represent the actual environmental conditions perceived by the home appliances. For example, regarding temperature, the probability distribution can represent the likelihood of different temperature values ​​occurring in a room, such as being closer to 25 degrees Celsius during the day and closer to 20 degrees Celsius at night. For humidity, the probability distribution may reflect the probability of different humidity levels occurring in a room. For light intensity, the probability distribution may represent the probability of changes in light intensity in a room at different times of the day. In practical applications, these probability distributions help home appliances understand and adapt to their environment. For example, if a home appliance knows that a certain temperature range (such as 20 to 25 degrees Celsius) is more likely to occur, it can more effectively adjust the operating mode of air conditioners or heating equipment based on this information.

[0051] In this context, the aforementioned environmental entropy represents the "uncertainty" of a home appliance's perception of the environmental state before it senses data. If the appliance can accurately predict environmental parameters (i.e., the probability distribution is relatively concentrated and the variation is small), the entropy value is low, indicating that the appliance's understanding of the environment is more certain and it can react more effectively. In summary, the probability distribution of environmental parameters quantifies the uncertainty of smart home appliances when facing environmental changes, which directly affects their ability to react and adapt to these changes.

[0052] After collecting the environmental information, a first conditional entropy is defined to describe the uncertainty of home appliances in response to environmental conditions after observing the sensor output. The formula for the first conditional entropy is: .

[0053] in, This indicates the probability of observing the sensor's output. Indicates the output of a given sensor The conditional probability of the environmental parameters. In this formula, It is the output result of the sensor of the home appliance. It represents a specific environmental signal measured by the sensor, such as temperature, humidity, light, etc. This indicates that the sensor outputs a specific value. The probability. For example, if the output value of the temperature sensor is =22 degrees Celsius, then it refers to the probability that the sensor will observe this temperature value within a certain period of time.

[0054] In real-world applications, the probability of the output This reflects the degree to which home appliances respond to environmental conditions and the reliability of environmental data. If a sensor's output value is common (for example, a temperature sensor that consistently outputs values ​​close to 22 degrees Celsius in a given area), then... A higher value indicates that the output value is relatively common and stable.

[0055] In home appliances, sensors generate output values ​​based on different environmental conditions. For example, a temperature sensor may output various temperature values, and the frequency of these output values ​​is reflected in this parameter. If the sensor outputs values ​​between 22 and 24 degrees Celsius most of the time, then these output values ​​are more likely to occur.

[0056] Output probabilities not only reflect the frequency of environmental conditions observed by home appliances, but may also reveal the stability or variability of the environment. For example, if a temperature sensor frequently outputs temperatures between 20 and 25 degrees Celsius under certain conditions, but rarely outputs these values ​​under other conditions, it can be inferred that the temperature fluctuations in the environment where the home appliance is located are relatively small.

[0057] The probability of this output can help home appliances determine the stability of the current environment and the validity of environmental information. In smart home appliance applications, if the output value of a certain sensor is rare, it may indicate that the sensor has malfunctioned or that the environment has undergone extreme changes.

[0058] Within the framework of information theory, this parameter not only helps quantify the frequency and reliability of sensor measurements but is also related to the device's decision-making process in dynamic environments. Smart home appliances can adjust their behavior based on the probability of these outputs; for example, they can automatically adjust their operating mode when a certain output value is frequently detected, or issue warnings or make adaptive adjustments when rare output values ​​occur.

[0059] In practical applications, by calculating the above entropy value, smart home appliances can respond efficiently in changing environments. By calculating the first conditional entropy, it is possible to quantify how much uncertainty about the environment remains after the home appliance is set to sense the environmental state.

[0060] Mutual information can be used to measure the information shared by two random variables. Knowing random variable X reduces the uncertainty of random variable Y (or knowing random variable Y reduces the uncertainty of random variable X). Therefore, in this application embodiment, a first environmental perception capability for home appliances is defined based on the environmental entropy and the first conditional entropy calculated above. The first environmental perception capability can be directly represented using the first mutual information based on the environmental state, and its formula is: .

[0061] I() represents the functional expression of mutual information, when the first mutual information... The higher the value, the more valuable the environmental information obtained by the home appliances, and the more effectively they can reduce uncertainty.

[0062] In one specific implementation, in order to perform standardized evaluation of different home appliances with different sensor configurations, the embodiments of this application can also perform normalization processing on the above-mentioned first mutual information, and the processing formula is as follows: .

[0063] After normalization, a value close to 1 indicates that the sensor is very effective in capturing environmental information, while a value of 0 indicates that the sensor has not captured meaningful environmental data.

[0064] Furthermore, after acquiring and digitizing environmental information, home appliances must internally encode and store the data in a compact yet information-rich manner. This process ensures the preservation of critical environmental signals while eliminating redundancy and noise. This phase involves balancing three core objectives: preserving critical environmental signals so that downstream reasoning and decision-making can operate with minimal uncertainty; compressing raw data into an internal representation that requires less computational and memory resources while maintaining the integrity of the device's understanding of its environment; and accurately simulating the probabilistic structure of the environment to support reliable prediction, pattern recognition, and adaptive updates in response to changing conditions.

[0065] like Figure 3 As shown, this core dimension quantifies the encoding effectiveness and utilization efficiency of environmental data by home appliances. Figure 1 Step S120 further includes steps S310 to S330: Step S310: Read the encoding structure of the home appliance in the first data processing loop and the original features obtained by the sensor sensing the environmental state. Based on the difference between the original features and the encoded data under the encoding structure, determine the encoding distortion of the home appliance for the environmental state information. Step S320: Based on the original features and coding structure, evaluate the authenticity retention of the coding structure of the home appliance during the first data processing loop, and determine the coding entropy difference; Step S330: Based on the difference in coding distortion and coding entropy, analyze the efficiency of home appliances in utilizing environmental status information and generate the first assessment report.

[0066] Consider the environment state E described by a random vector and its underlying reference distribution. After initial perception and preprocessing, the device constructs an encoding structure I to encode key environmental features. Here, the encoding process is defined as I processing the data S obtained from the sensor through a transformation function g(). This transformation function can be represented as a compression algorithm, statistical model, or embedded machine learning. The transformation function needs to parse the environmental information S while minimizing redundancy.

[0067] In this embodiment, the reference distribution of the original features is first estimated based on actual environmental measurements. The reference distribution reflects the spatial complexity of the actual environmental state. During the process of collecting environmental information, due to the limitations of machine coding, the encoding of environmental information by home appliances may lose the authenticity of the original signal. Therefore, the internal model distribution is derived from the coding structure of home appliances. .

[0068] Similar to the approach using the perceived environmental state dimension, data discretization and probability estimation are necessary. Continuous variables are partitioned into discrete states, and a large sample set is used to construct stable probability estimates. Noise modeling and smoothing methods are employed to ensure robust distribution approximations.

[0069] In reality, to obtain an accurate reference distribution, it is necessary to accurately represent it internally within the machine. A large amount of real-world data is required. Embodiments of this application may collect usage logs over a representative period and apply robust statistical methods (such as kernel density estimation or histogram binning with smoothing) to estimate these distributions. Sensitivity analysis should be performed to assess the impact of estimation errors.

[0070] The distribution results of the model represented by the encoding structure Reference distribution corresponding to the real environment By comparing the results, we can obtain the encoding distortion of environmental information within the home appliances. Here, the encoding distortion is quantified using KL divergence. .

[0071] and The encoding distortion between the model distribution and the reference distribution measures the difference in environmental information, thus measuring the model distribution represented by the encoding structure. To approximate the reference distribution corresponding to the real environment The amount of information lost, and the lower coding distortion, indicate that the encoding of environmental information inside the home appliance can approximate the real environmental state, thereby reducing the calculation error of the model.

[0072] Furthermore, this application also proposes an evaluation of the authenticity retention of environmental information using encoding based on the concept of mutual information. First, considering the original features R of the real environmental state and its encoded data I, the entropy value of the original features R is calculated. and the conditional entropy of encoded data based on the original features The difference between the entropy values ​​after encoding is calculated to obtain the encoding entropy difference based on mutual information representation: .

[0073] In one specific implementation, in order to standardize the expression of numerical values, as described above, Entropy representing the original features The difference represents the entropy of the internal coding structure. - This indicates that the encoded features have removed redundant or noisy elements from the environmental state. Here, the difference in encoded entropy is normalized. It can express the retention of original state information. When it is close to 1, it means that most of the environment-related information is retained.

[0074] In order to generate a single evaluation value based on the encoded features, embodiments of this application may further weight and mix the above data: .

[0075] in, A standardized weighting factor representing the relative importance of the fidelity of encoded features to their encoding efficiency. .

[0076] It is a standardized constant representing the worst-case divergent scenario. In actual implementation, The weighted mixture should be defined based on the worst-case scenario observed in the preliminary experiments. Scaling to the range [0,1], close to 1 This represents an efficient and accurate representation of internal coding features. By selecting appropriate normalization and weighting schemes, this single score can be directly plotted on the radar axis to indicate the device's performance in coding and representation.

[0077] high The scores indicate that the appliance successfully converts raw, potentially noisy sensor readings into a stable, low-entropy internally encoded representation that closely resembles the real environment and preserves essential information. Because its internal state is accurate, compact, and easy to use, the appliance is able to improve reasoning and decision-making.

[0078] Conversely, lower scores This could indicate that the device's internal model is inconsistent with reality, potentially leading to incorrect predictions or suboptimal behavior. It could also mean that excessive compression discards too much information, impairing the device's ability to infer future states or adapt to new conditions.

[0079] It is worth noting that in practical applications, changes in environmental conditions and usage patterns can affect these scores. Therefore, it is recommended to recalibrate the metrics periodically and validate them against independent performance benchmarks to maintain accuracy.

[0080] Because these metrics rely on probability distributions and established information theory measures, the evaluations are objective and repeatable. The same metrics can be used to compare different devices with different encoding strategies and memory architectures. This allows evaluators to identify bottlenecks, improve compression algorithms, or refine model training procedures.

[0081] After acquiring environmental data and converting it into a valid internally encoded feature representation, home appliances must accurately infer hidden or future states and choose appropriate actions to achieve their goals or tasks under uncertainty. Reasoning involves reasoning about the hidden or future states of the environment (e.g., predicting how room temperature might change in the next hour), while decision-making involves choosing an action from possible alternatives under uncertainty (e.g., adjusting cooling power or scheduling cleaning cycles). This core dimension details the theoretical foundations of these processes and introduces information theory measurement methods that provide normalized scores for the "Reasoning and Decision-Making under Uncertainty" dimension.

[0082] Specifically, such as Figure 4 As shown, Figure 1 Step S130 includes, but is not limited to, steps S410 to S430: Step S410: Read the set of action instructions that the home appliance infers based on its internal knowledge after receiving environmental data, evaluate the home appliance's ability to control the environment under the influence of internal knowledge based on the set of action instructions, and determine the second conditional entropy. Step S420: Map the environmental data to the action command set, so that the home appliance selects and executes the environmental control action based on the action command set, obtains the expected environmental state predicted by the home appliance based on internal knowledge, and determines the environmental state difference based on the set target environmental state and the expected environmental state. Step S430: Evaluate the ability to acquire internal knowledge and the performance of home appliances based on the second conditional entropy and the difference in environmental state, and generate a second evaluation report.

[0083] Reasoning and decision-making constitute the cognitive core of smart home appliances, including predicting future or latent environmental states and choosing the best action in uncertain situations. Home appliances can maintain probabilistic models of potential states (e.g., upcoming temperature patterns, occupant behavior) and aim to select action A that optimizes certain objectives (e.g., minimizing energy consumption or maintaining user comfort). Numerically, based on the above concept, let: I represents the encoding feature mentioned earlier, indicating processed sensor data such as temperature, humidity, and light intensity.

[0084] Z represents unobserved or future environmental conditions; This refers to a set of action instructions that a home appliance can take based on acquired environmental information.

[0085] After encoding the environmental information collected by home appliances into a feature representation I, information theory indicators are applied to infer the second environmental change process and generate an environmental change probability distribution. The process of this inference output distribution can be derived from the following formula: .

[0086] The second conditional entropy is calculated based on the probability distribution of environmental changes obtained through reasoning. It is used to measure the uncertainty in reasoning about environmental changes. The second conditional entropy is specifically expressed as follows: .

[0087] Here, s, z, and i are elements in S, Z, and I, respectively, and a low value of the second conditional entropy indicates that the home appliance can accurately predict unobserved environmental states through encoded features.

[0088] Here, based on the concept of mutual information, the correlation between the two sets of variables, namely the internal coding features I of real-time environmental information and the unobserved environmental state Z, is quantified again. Specifically, the formula is: .

[0089] In this formula, The entropy represents the unobserved environmental state Z, indicating the uncertainty of the environmental state when there is no environmental information. Given data I, the conditional entropy of environmental state Z represents the remaining uncertainty about environmental state Z after the home appliance has collected data I. Mutual information is also relevant. This represents the reduction in uncertainty about environmental state Z through data I. The higher the value, the stronger the predictive power of data I for environmental state Z and the more information it provides.

[0090] In addition, regarding the set of action instructions inside home appliances, once the home appliance selects an action instruction... The resulting distribution then becomes Assuming action instruction 'a' represents a specific cooling strategy, It can represent room temperature and energy usage over a period of time. Accumulating this data allows for... Make an estimate.

[0091] Specifically, the instruction distribution after selecting an action command represents the probability distribution of the expected results or consequences of the environment or system after a home appliance selects a specific action 'a'. This describes the possible results that an action will bring, and the probability of these results occurring; while It refers to the expected future environmental or system state after the action is performed, such as the target indoor temperature after the air conditioner lowers the temperature.

[0092] Taking a smart air conditioner as an example, suppose the air conditioner decides whether to turn on the cooling function based on the current room temperature (I sensed by a sensor). The air conditioner might choose to "turn on cooling" or, based on the user's needs, select "keep the temperature constant." After the air conditioner decides to "turn on cooling," the change in indoor temperature is an uncertain process. The change in indoor temperature depends not only on the air conditioner settings but also on various factors such as room insulation performance, external climate, and the number of people. It describes the multiple possible states of indoor temperature after the "cooling" action is activated, along with their probabilities. For example, after the air conditioner is turned on for cooling, the temperature may drop to 22 degrees Celsius (60% probability), 20 degrees Celsius (30% probability), or 18 degrees Celsius (10% probability). This probability distribution reflects the uncertainty of the environment.

[0093] In general, outcome distribution refers to the distribution of possible outcomes and their probabilities after an intelligent device (such as an air conditioner or robot) performs a certain action. It reveals the uncertainty following the device's decision and reflects the uncertainty of the environment or system response. This distribution helps the device make better decisions because it quantifies the probability of each possible outcome and selects the optimal action amidst uncertainty.

[0094] In this core dimension, actions driven by motion commands are mapped to expected environmental states. Once the device selects action 'a', an expected environmental state will be generated. Based on the concept of mutual information, the expected environmental state is calculated after selecting action a. Impact effects: .

[0095] Here, if A large value indicates that the implementation of the set of action instructions A for the home appliances will significantly reduce the uncertainty of the environmental state and can guide the environmental state to the ideal state in a timely manner.

[0096] In most cases, home appliances are designed to maximize the utility function. This represents the desired environmental control outcome, such as a comfortable temperature range, minimum power consumption, energy efficiency, user comfort, or other performance standards. These standards are defined based on specific appliance objectives. The appliance uses a utility-maximizing function combined with the aforementioned mutual information to form a comprehensive evaluation result of the set of action commands. This comprehensive evaluation result can be specifically expressed as... Where E() represents the expected value, and A0 represents the random or static strategy, then This represents the expected environmental state under random and static policies.

[0097] To standardize the above evaluation information, integrate the above reasoning and decision-making, and ensure the accuracy of reasoning and the effectiveness of decision-making, the same calculation index is also given here: .

[0098] in, Indicates the weighting factor. Used to assess the accuracy of home appliances' internal knowledge in predicting unknown environmental conditions. Used to measure the effectiveness of selected action instructions in influencing or optimizing future environmental conditions.

[0099] It should be noted that the internal knowledge of home appliances includes, but is not limited to, pre-set linear instruction sets, statistical models, pre-trained large models, or hybrid decision modules that combine multiple decision types. This internal knowledge is involved in tasks such as organizing environmental information, analyzing environmental conditions, and generating and calling action instructions.

[0100] High values This indicates that the device can reliably predict environmental or user conditions and select operations that will produce better results. Conversely, low values... This may indicate that home appliances are unable to accurately predict future environmental conditions, or that the set of action instructions for home appliances cannot adequately optimize the current environmental conditions.

[0101] By employing these standardized, probabilistic measures, evaluation results remain objective, repeatable, and comparable across various product categories and operating environments. Suboptimal devices can be identified and improved by refining inference algorithms (e.g., Bayesian updates, machine learning models) or adjusting action selection strategies (e.g., reinforcement learning, rule-based optimization).

[0102] Calculating this index requires estimating the probability distribution based on actual operational data. However, the real-world environment is dynamic, and the probability distribution used here may change over time or with variations in the environment. Furthermore, the device may continuously learn from observed user interactions and environmental feedback to update its internal knowledge. Embodiments of this application may employ empirical sampling techniques, real-world experiments, or validated simulation models to reliably acquire these parameters.

[0103] Intelligent home appliances must not only sense their environment and make informed decisions, but also continuously adapt their internal knowledge to changing conditions or emerging data, thereby achieving sustained and improving performance. This adaptive learning capability distinguishes truly intelligent systems from static, pre-programmed devices. Whether the device is dealing with evolving user preferences, seasonal changes in environmental parameters, or new data streams from external sources, continuous learning enables more accurate reasoning, better decision-making, and consistent performance.

[0104] Therefore, embodiments of this application will also provide how an appliance can quantify its learning and adaptation process using information-theoretic metrics. Similar to the previous dimensions, the ultimate goal is to derive a single numerical value that reflects the appliance's ability to reduce uncertainty by repeatedly observing and updating its internal knowledge representation under changing conditions.

[0105] like Figure 5 As shown, Figure 1 Step S130 further includes, but is not limited to, the following steps S510 to S520: Step S510: The home appliance acquires initial environmental information and controlled environmental information after completing the environmental control action, so that the home appliance updates its internal knowledge based on the initial environmental information and controlled environmental information. Step S520: Evaluate the update level of internal knowledge, determine the knowledge gain of internal knowledge, and generate a second evaluation report.

[0106] It should be noted that, in this embodiment, internal knowledge specifically refers to the module integrated into the home appliance that is used to analyze data and participate in decision-making.

[0107] In an adaptive learning environment, the device maintains a dynamically evolving internal model that represents its behavior over time. The model is based on a current understanding of the environmental state or user preferences. It is updated periodically to incorporate new information and improve predictive accuracy. This is done while acquiring new data. After receiving (e.g., updated sensor readings, user feedback), the device will change its model from... Refined to Optionally, the internal knowledge is updated in a Bayesian manner, and the adaptive learning process of the internal knowledge is shown in the following equation: .

[0108] The normalized Bayesian update equation clearly illustrates the new data. Update the device's internal knowledge for .

[0109] From an information theory perspective, adaptive learning can be viewed as a process of reducing entropy or minimizing divergence over time. Through the aforementioned Bayesian update, the internal knowledge of home appliances will steadily reduce the gap between its internal model and the true distribution of the environmental state. This application's embodiments will be designed based on the response efficiency of home appliances updating with new data, the ability of each update to improve the sensing effect of the environment (i.e., eliminate uncertainty), and whether the internal knowledge can converge to an accurate description of the environment or user preferences within a finite number of updates.

[0110] To ensure computational feasibility, the internal knowledge should be explicitly discretized or parameterized (e.g., Gaussian mixture) at each iteration (time step). Explicitly defining the discretization or parameterization process helps obtain stable and reliable probabilistic estimates required for accurate entropy and divergence calculations; here, these continuous distributions are represented as time... The model, for each model update, the distribution This conclusion was drawn from newly observed data.

[0111] Furthermore, adaptive learning typically utilizes multiple data sources: sensor readings, user interactions, or external knowledge. Each update step aggregates these sources into an updated model distribution. The resulting probability distribution must be normalized and smoothed or regularized to handle zero or low-frequency intervals.

[0112] Furthermore, if real data or a well-established baseline distribution is available, the new model can be benchmarked to measure improvements or persistent inaccuracies.

[0113] To quantify the adaptive learning process, it is necessary to calculate the learning optimization efficiency of internal knowledge, which can be defined using KL divergence: .

[0114] In a progressive update process, a high learning optimization efficiency indicates a substantial model update, meaning newly observed data significantly alters the device's understanding of the environment or user behavior. Conversely, low divergence indicates minimal novelty in recent data, or that the existing model already fits the environment well. Low divergence suggests the model is already accurate, or that new data provides minimal novelty. Learning optimization efficiency reveals the speed and stability of learning across multiple update cycles.

[0115] In another specific implementation, the reference distribution representing the real environmental state is referenced again. In the old model Update to the new model Afterwards, measurements can be taken before and after the update. and Has it converged to To determine whether the internal knowledge update has converged, under this theory, it can also be done using a formula. To quantify the efficiency of learning optimization.

[0116] Furthermore, this embodiment of the application quantifies the knowledge gain within each update process during the progressive update process, given new data. The old model of the time and new models The knowledge gain between them, where gain represents the effective reduction of entropy through internal knowledge updates, can be expressed by the following formula: ; .

[0117] Knowledge gain quantifies the effective use of new data, with a larger ΔH reflecting better learning achieved through updates and reduced uncertainty in internal knowledge's perception of the environment.

[0118] In real-world scenarios, the precise probability distributions required for entropy and divergence calculations may not be readily available. To overcome this problem, home appliances can reliably estimate these distributions from limited datasets using empirical sampling methods, statistical inference, or approximation techniques such as particle filters and histogram estimation. This ensures that the metrics proposed in the embodiments of this application are not merely theoretical but can be implemented in practice, even under conditions of limited computational or memory resources.

[0119] Comprehensive Learning Score To integrate the numerical values ​​of learning optimization efficiency and knowledge gain, the comprehensive learning score is defined as follows: .

[0120] in, This represents the average value of ΔH within the preset update time range. This represents the residual of learning and optimization efficiency within a preset update time range. This indicates the preset weighting factor.

[0121] Optionally, in order to provide standardized evaluation metrics, embodiments of this application also propose a definition of a normalized comprehensive learning score: .

[0122] In this formula, This represents the theoretically maximum possible knowledge gain. This indicates the maximum efficiency of learning optimization. This is a preset weighting factor.

[0123] In summary, the comprehensive learning score, which combines learning optimization efficiency and knowledge gain, reflects how much the internal knowledge of home appliances has improved over time and how closely it approximates real-world behavior in the environment. High scores... It clearly demonstrates effective model adaptation and learning across multiple data updates, showing that the device can reliably improve its internal knowledge over time.

[0124] In the modern smart home ecosystem, home appliances rarely operate independently. Instead, they rely on network connectivity to exchange data with other devices, access cloud services, or utilize external knowledge bases. Therefore, embodiments of this application also evaluate the ability of home appliances to effectively utilize interconnectivity and external resources to enhance their internal intelligence. By integrating additional data and computing power beyond onboard resources into their own systems, appliances can reduce uncertainty, improve models, and make more informed decisions.

[0125] The final result of this core dimension should also be a single numerical value to reflect the quality and impact of external integration of home appliances. This value will be combined with the other four dimensions in the radar chart to form a comprehensive multi-axis intelligence assessment.

[0126] like Figure 6 As shown, in some embodiments of this application, the evaluation method proposed in this application further includes the following steps S610 to S630: Step S610: Simulate the third environmental change process and control the home appliance group to adjust to the third environmental change process. The home appliance group includes multiple home appliances that are interconnected and are set up in the same test space. Step S620: Collect the second data processing loop within the home appliance group and detect the interconnection status within the home appliance group; Step S630: Generate a third assessment report based on the interconnection status.

[0127] It should be noted that external entities that enable interconnection within the home appliance group include, but are not limited to: cloud platforms: offloading data storage or utilizing cloud-based AI / ML services; other smart devices: exchanging data point-to-point or hub-mediated with thermostats, cameras, or voice assistants; user devices: mobile phones, tablets, or wearable devices that provide user feedback or control signals; and IoT gateways: routers or dedicated smart home hubs that provide advanced coordination functions.

[0128] External resource utilization focuses on how these connections improve device performance. For example, home appliances can retrieve weather forecasts from the cloud or utilize centralized AI models trained on aggregated user data. The more effectively devices integrate this external information, the more they can reduce uncertainty, adjust decisions, and enhance the learning process.

[0129] To quantify the extent and quality of external utilization, this core dimension also needs to capture relevant probability distributions to describe the external data. How to determine the encoding features or expected result distribution within home appliances? Make changes.

[0130] Specifically, firstly, external data is constructed. probability distribution Then, conditional probability distributions are constructed respectively. and It is used to assess how the coding characteristics and expected outcome distribution within home appliances change after receiving external data.

[0131] During the interconnection and interoperability process of home appliance groups, the data interaction efficiency within the home appliance group is evaluated. For a home appliance in the group that intends to perform calculations, the data throughput of that home appliance is assessed. Quantization, data throughput The data interaction efficiency is calculated by dividing the total amount of data transmitted and processed by the total time spent, and then obtaining the channel capacity C for communication between the home appliance and the outside world. .

[0132] Using mutual information theory again, we first calculate the entropy of the first environment information utilization efficiency under the influence of only the coding features within a single household appliance during the first environmental change process. In the process of change in the third environment of interconnection, the entropy of the uncertainty represented by the information utilization efficiency of the second environment through the influence of home appliance groups on the internal representation of the device. This is expressed as follows: The first group effect is determined by the efficiency of utilizing first and second environmental information. .

[0133] In addition, to assess how external inputs improve the accuracy of predicting future environmental conditions, the initial environmental regulation effect of a single household appliance was obtained after the second environmental change process. By gradually adding external devices with communication connections to home appliances, a third environmental change process is carried out sequentially to obtain the second adjustment effect of the environment. Then, the theory of mutual information is used to judge the second group effect of the interconnection process on the environmental optimization effect of home appliances: .

[0134] To assess the average impact of adding external devices on home appliances, the following definition is provided. The mean of the second group effect during the gradual addition of external devices. This is the second group effect after adding all theoretically possible external devices.

[0135] In order to standardize the evaluation scores for interoperability, this application defines the following in its embodiments: .

[0136] Here, All are weighting factors. .

[0137] High values This demonstrates that the device effectively utilizes external data and communication to enhance its intelligence. Conversely, when A low score indicates limited or ineffective connectivity usage, failing to leverage the potential benefits of resource sharing and cloud-based analytics.

[0138] Based on the above five core dimensions, the embodiments of this application propose to include... , The first assessment report, containing , The second assessment report, and containing The third assessment report, which includes information collection and physical sensing ( Assess how home appliances can effectively reduce environmental uncertainty through accurate and reliable sensing, information encoding, storage, and internal representation. ) Evaluate the efficiency and fidelity of coding and retaining basic information of home appliances, and make decisions under uncertainty in reasoning and decision-making. ) measures the ability of home appliances to predict future states and make optimal decisions based on uncertain information, adaptive information updates and learning. Quantify the ability of home appliances to continuously improve their internal models based on new data and changing conditions, including interconnectivity and external resource utilization. This measures how effectively home appliances utilize external data sources and computing resources to enhance their intelligence.

[0139] A high score in any single dimension does not necessarily guarantee superior overall intelligence. For example, while mutual information measures an appliance's ability to capture relevant environmental signals, it may not encompass its ability to make timely or optimal decisions. Therefore, by combining a comprehensive evaluation of five core dimensions and generating a five-dimensional radar chart from the evaluation report produced by these five dimensions, the necessity of a multi-dimensional evaluation framework can be emphasized. The balanced performance of each dimension better reflects the overall intelligence level of the device.

[0140] Additionally, refer to Figure 7 This application also provides a smart performance evaluation device 700 for home appliances, comprising: The simulation module 701 is used to build a smart performance evaluation model for home appliances, simulate the first environmental change process, and initiate the home appliance to identify the first environmental change process so that the home appliance can acquire environmental data. The first evaluation module 702 is used to collect the first data processing cycle process in the home appliance, evaluate the first data processing cycle process, and generate a first evaluation report. The second assessment module 703 is used to enable home appliances to perform environmental control actions, identify the second environmental change process, evaluate the environmental control capabilities of home appliances based on the second environmental change process, and generate a second assessment report.

[0141] The aforementioned smart home appliance performance evaluation device 700 and the quantitative evaluation method for the intelligence level of multimodal smart home appliances are based on the same inventive concept, and will not be described in detail here.

[0142] Additionally, refer to Figure 8 , Figure 8 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 801 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 802 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 802 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 802 and is called and executed by the processor 801 to execute the quantitative evaluation method for the intelligence level of multimodal smart home appliances according to the embodiments of this application. The 803 input / output interface is used to implement information input and output. The communication interface 804 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 805 transmits information between various components of the device (e.g., processor 801, memory 802, input / output interface 803, and communication interface 804); The processor 801, memory 802, input / output interface 803, and communication interface 804 are connected to each other within the device via bus 805.

[0143] This application embodiment also provides a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs, which can be executed by one or more processors to realize the above-mentioned quantitative evaluation method for the intelligence level of multimodal smart home appliances.

[0144] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0145] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0146] It will be understood by those skilled in the art that Figures 1 to 6 The technical solutions shown do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

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

[0148] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0149] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0150] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0151] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above 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.

[0152] The units described above 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.

[0153] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0154] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0155] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A quantitative evaluation method for the intelligence level of multimodal smart home appliances, characterized in that, include: A smart performance evaluation model for home appliances is constructed to simulate a first environmental change process. The home appliances are then activated to identify the first environmental change process so that they can acquire environmental data. The first data processing loop within the home appliance is collected, the first data processing loop is evaluated, and a first evaluation report is generated. The home appliance is made to perform environmental control actions, a second environmental change process is identified, the environmental control capability of the home appliance is evaluated based on the second environmental change process, and a second evaluation report is generated. The step of collecting the first data processing loop within the home appliance, evaluating the first data processing loop, and generating a first evaluation report includes: The data types and corresponding environmental data acquired by the home appliance during the first data processing loop are read. A data type matrix is ​​constructed based on the acquired data types. The ability of the home appliance to identify environmental states is evaluated based on the data type matrix, and the unknown environmental states before the home appliance perceives the environmental data are solved to obtain environmental entropy. An environmental information matrix is ​​constructed based on the environmental data. The ability of the home appliance to perceive environmental states is evaluated again based on the environmental information matrix to obtain a first conditional entropy. Based on the environmental entropy and the first conditional entropy, the ability of the home appliance to perceive the unknown environmental states using the environmental information matrix is ​​judged to determine the first environmental perception capability of the home appliance. The system reads the encoding structure of the home appliance during the first data processing loop, as well as the original features obtained from the environmental state perceived by the sensors. Based on the difference between the original features and the encoded data under the encoding structure, it determines the encoding distortion degree of the home appliance for environmental state information. Based on the original features and the encoding structure, it evaluates the authenticity retention of the encoding structure of the home appliance during the first data processing loop and determines the encoding entropy difference. Based on the encoding distortion degree and the encoding entropy difference, it analyzes the utilization efficiency of the home appliance for environmental state information. Based on the first environmental perception capability and the utilization efficiency of the home appliance for environmental state information, it generates a first evaluation report. The step of causing the home appliance to perform environmental control actions, identifying a second environmental change process, evaluating the home appliance's environmental control capabilities based on the second environmental change process, and generating a second evaluation report includes: After receiving environmental data, the home appliance infers a set of action commands based on its internal knowledge. Based on this set of action commands, the home appliance's ability to manipulate the environment under the influence of the internal knowledge is evaluated, and a second conditional entropy is determined. The environmental data is mapped to the set of action commands, allowing the home appliance to select and execute environmental control actions based on the set of action commands. The expected environmental state predicted by the home appliance based on the internal knowledge is obtained. An environmental state difference is determined based on the set target environmental state and the expected environmental state. The capabilities of the internal knowledge and the performance of the home appliance are evaluated based on the second conditional entropy and the environmental state difference. The home appliance acquires initial environmental information and regulated environmental information after completing environmental regulation actions, enabling it to update its internal knowledge based on these information. This internal knowledge is updated gradually over a period of time, with the discrete parameters of the internal knowledge changing before and after each update node. The convergence of the internal knowledge is determined based on these parameter changes, and the learning optimization efficiency is assessed. The knowledge gain between adjacent update nodes is determined. Finally, a second evaluation report is generated based on the assessment of the internal knowledge's capabilities, the home appliance's performance, the learning optimization efficiency, and the knowledge gain.

2. The quantitative evaluation method for the intelligence level of multimodal smart home appliances according to claim 1, characterized in that, The method further includes: The process of simulating a third environmental change is used to control a group of home appliances to adjust to the third environmental change process. The group of home appliances includes multiple home appliances that are interconnected and are set up in the same test space. The second data processing loop within the home appliance group is collected to detect the interconnection status within the home appliance group; A third assessment report is generated based on the aforementioned interconnectivity situation.

3. The quantitative evaluation method for the intelligence level of multimodal smart home appliances according to claim 2, characterized in that, The generation of the third assessment report based on the interconnection status includes: During the interconnection and interoperability process of the home appliance group, the data interaction efficiency of the home appliance group is collected; During the first environmental change process, the first environmental information utilization efficiency of the home appliances is obtained. During the third environmental change process, environmental information is transmitted between the home appliances, and the second environmental information utilization efficiency of the home appliances is collected. The first group effect is determined based on the first environmental information utilization efficiency and the second environmental information utilization efficiency. After the second environmental change process, the first adjustment effect of the home appliances on the environment is detected; after the third environmental change process, the second adjustment effect of the home appliance group on the environment is detected; and the second group effect is determined based on the first adjustment effect and the second adjustment effect. A third evaluation report is generated based on the data interaction efficiency, the first group effect, and the second group effect.

4. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the quantitative evaluation method for the intelligence level of multimodal smart home appliances as described in any one of claims 1 to 3.

5. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the quantitative evaluation method for the intelligence level of multimodal smart home appliances as described in any one of claims 1 to 3.

Images