SYSTEM AND METHOD FOR DETECTING ANOMALIES

Description

Title System and method for detecting anomalies System and method for detecting anomalies DESCRIPTION Technical scope The present invention relates to a home automation system and a method for detecting anomalies. Technological background The invention finds particular, though not exclusive, application in the technical sector pertaining to home automation. Nowadays, there is an ever-increasing need to monitor the use of electrical and electronic devices installed in a home automation system. This is to meet various needs, such as energy saving, simplifying home life, and personal safety. There is also a need to monitor the habitual behavior of home automation system users so as to detect any deviations that could pose potential dangers to the user and / or to improve the user's life management. An example of a monitoring system using purpose-built sensors is described in U.S. Patent No. US 10,609,342 B1. However, the Applicant has noted that the current technical solutions for the monitoring described above make the home automation system of a home, or more generally, a building, architecturally more complex. Furthermore, these solutions inevitably increase the installation and maintenance costs of the home automation system. The Applicant therefore understood the opportunity and need to detect anomalies in the use of a home automation system through relatively inexpensive devices to produce. Furthermore, the Applicant understood the usefulness of detecting anomalies without excessively aggravating the complexity of the home automation system. The technical problem underlying the present invention is therefore that of providing a system and a method for detecting structural and functional anomalies designed to at least partially overcome one or more of the drawbacks complained of with reference to the cited prior art. In the context of this problem, it is an object of the present invention to provide a system and method for detecting anomalies that is relatively inexpensive to implement. A further aim of the present invention is to provide a system and method for detecting anomalies that does not make a home automation system already installed in a building more complex, particularly from a construction point of view. This problem is solved and at least one of these objects is at least partially achieved by the invention by means of a system and a method for detecting anomalies according to one or more of the respective appended claims. In this description, as well as in the claims attached thereto, certain terms and expressions are deemed to have, unless otherwise explicitly indicated, the meaning expressed in the definitions that follow. The term "electrical appliance" refers to one or more electrical devices and / or one or more electronic devices. Preferably, said one or more electrical devices and / or one or more electronic devices are present in a home or, more generally, in a building, and are designed to perform a specific function. The term anomaly refers to a change or deviation in the data indicative of an operating state assumed by an electrical appliance from a set of predetermined data. By the term objects we mean, preferably, human beings or animals. Summary of the invention In its first aspect, the present invention is directed to a system for detecting anomalies, in particular in the use of a home automation system. The system comprises at least one electronic device operatively connected to a communications network and to an electrical appliance of a home automation system, the electrical appliance being configured to assume a plurality of predefined operating states. Preferably, the system also comprises the home automation system. The home automation system may be installed in a domestic or non-domestic building. Said at least one electronic device is configured to transmit data signals over the communications network including data indicative of an operating state assumed by the electrical appliance among the aforementioned plurality of predefined operating states. Preferably, the data signal includes additional information in addition to the data indicating the operating state assumed by the electrical appliance. Specifically, the additional information may include data representing the point in time (e.g., date and / or time) at which the electrical appliance assumed that operating state. More preferably, the electrical appliance can assume at least two predefined operating states, namely a first operating state (e.g. on, open or active) and a second operating state (e.g. off, closed or inactive). Preferably, the electrical appliance is configured to transmit information about its operating status to the at least one electronic device (e.g. via the aforementioned communication network or over the network) and / or the at least one electronic device is configured to read the operating status of the electrical appliance (e.g. via the aforementioned communication network or over the network). Preferably, said at least one electronic device comprises a control device configured to control the operation of the electrical appliance (in particular for at least one of: activating, operating, deactivating, switching on, switching off and regulating the electrical appliance), more preferably via the communication network. The system comprises a data processing device operatively connected to the electronic device via the communications network and configured to receive data signals transmitted by the electronic device. The data processing device includes a module configured to detect an anomaly in the data signals received from at least one electronic device and to generate an output signal indicative of the presence of said anomaly based on said detection. Thanks to these features, the system allows for the detection of anomalies in the use of the home automation system by a user (intended as one or more people), particularly in the use of an electrical appliance, and, consequently, to recognize a significant deviation in the user's behavior from a normal pattern. Furthermore, the use of data indicative of the operating status of the electrical appliance advantageously allows the information normally circulating in the home automation system to be exploited to detect anomalies without the burden of using specific sensors. The result is a system according to the invention that is relatively simple and inexpensive to implement. Furthermore, if at least one electronic device includes the control device, a single device would be obtained that could both control the electrical appliance and detect its status. This advantageously avoids the need to install excessive physical devices in the home automation system. In its second aspect, the present invention is directed to a method for detecting anomalies. The method includes: • operatively connecting an electronic device to a communication network of a home automation system, • operatively connecting the electronic device to an electrical appliance of the home automation system, the electrical appliance being configured to assume a plurality of predefined operating states and said electronic device being configured to transmit data signals into the communication network comprising data indicative of an operating state assumed by the electrical appliance among said plurality of predefined operating states, • operatively connecting a data processing device to the electronic device via the communication network, wherein the data processing device comprises a module configured to detect an anomaly in the data signals received from the electronic device, • transmitting data signals into the communication network by means of the electronic device,• receive in the module the data signals transmitted by the electronic device, • detect by means of the module whether an anomaly is present in the received data signals, • generate by means of the module an output signal indicative of the presence of said anomaly on the basis of said detection. Specifically, the method may include a module evaluation step based on the received data signals. This module evaluation step may include the module detecting whether an anomaly exists in the received data signals. The aforementioned evaluation phase preferably involves executing the module based on the data signals transmitted by the electronic device. The above method therefore allows for the detection of anomalies in the use of the home automation system. In at least one of the above aspects, the present invention may further exhibit at least one of the preferred features set forth below. In at least one embodiment of the present invention, the module comprises one of: • a machine learning model trained to detect said anomaly in data signals received from said at least one electronic device, and • a probabilistic algorithm configured to detect said anomaly in data signals received from said at least one electronic device based on at least a comparison of the probability of said electrical appliance assuming the operating state represented in said data signals with a predetermined threshold. The machine learning model is preferably configured to generate an output signal indicative of the presence of said anomaly based on said detection. Preferably, said anomaly is detected by the machine learning model if the data signals received from the at least one electronic device differ from those predicted by the machine learning model based on its learning, preferably over a predefined period of time. The probabilistic algorithm is preferably configured to generate an output signal indicative of the presence of said anomaly based on said detection. In at least one embodiment of the present invention, the machine learning model is subjected to a training phase aimed at training the model to detect said anomaly. Preferably, the data sent as input to the model for the training phase correspond to a set of data signals transmitted by the at least one electronic device within a first pre-established time interval, wherein the set of data signals comprises data indicative of the operating states assumed by the electrical appliance in the first pre-established time interval. In at least one embodiment of the present invention, the machine learning model is trained to detect an anomaly in the data signals received from the at least one electronic device within a second predefined time interval. Preferably, the second predefined time interval is shorter than the first predefined time interval. In at least one embodiment of the present invention, the machine learning model is configured to generate said output signal indicative of the presence of an anomaly within the second predefined time interval if: • said anomaly is detected by said machine learning model in one of the data signals received in the second predefined time slot, or • said anomaly is detected by said machine learning model in all data signals received in the second predefined time slot, or • said anomaly is detected by said machine learning model in at least a predefined percentage of the data signals received in the second predefined time slot (e.g. in at least 50% + 1 of the data signals received in the second predefined time slot). In at least one embodiment of the present invention, the machine learning model comprises at least one of the following machine learning algorithms: Isolation Forest, Gaussian Mixture Model, One-Class SVM, and Elliptic Envelope. In addition to the machine learning model, the module may preferably implement statistical techniques, such as at least one of mean and variance, to process the information produced by the machine learning model to generate the aforementioned output signal. In at least one embodiment of the present invention, the module comprising the probabilistic algorithm is subjected to a training phase. Preferably, the data sent as input to the module for the training phase correspond to a set of data signals transmitted by the at least one electronic device within a first predetermined time interval. The set of data signals includes data indicative of the operating states assumed by the electrical appliance in the first pre-established time interval. For each operating state that can be assumed by the electrical appliance, the training phase allows the probabilistic algorithm to determine the probabilities that such operating state will be assumed by the electrical appliance at respective moments in time included in the first pre-established time interval (for example, for each minute of each day belonging to the first pre-established time interval) and, preferably, further probabilities that such operating state will be assumed by the electrical appliance at respective moments in time included in one or more different sub-intervals of the first pre-established time interval. A sub-interval can correspond to one of a season of the year (spring, summer, autumn, or winter), a month of the year (January, February, ..., November, and December), Mondays of the year, Tuesdays of the year, Wednesdays of the year, Thursdays of the year, Fridays of the year, Saturdays of the year, Sundays of the year, weekdays of the year, and holidays of the year. In at least one embodiment of the present invention, the probabilistic algorithm is configured to compare at least one probability that said electrical appliance will assume the operating state represented in a data signal received from the at least one electronic device with a predetermined threshold. Preferably, said at least one probability corresponds to the probability determined by the probabilistic algorithm in relation to the first pre-established time interval. Alternatively, said at least one probability may correspond to a plurality of probabilities determined by the probabilistic algorithm respectively in relation to the first pre-established time interval and to one or more different sub-intervals of the first pre-established time interval. Preferably, the probabilistic algorithm is configured to detect an anomaly in the received data signal if a predefined percentage of said at least one probability is less than, or equal to, the predefined threshold. More preferably, the probabilistic algorithm is configured to detect an anomaly if the majority of said at least one probability is less than, or equal to, the pre-specified threshold. In at least one embodiment of the present invention, the probabilistic algorithm is configured to generate said output signal indicative of the presence of an anomaly within a second predefined time interval if: • said anomaly is detected by said probabilistic algorithm in one of the data signals received within said second predefined time interval, or • said anomaly is detected by said probabilistic algorithm in all data signals received within the second predefined time interval, or • said anomaly is detected by said probabilistic algorithm in at least a predefined percentage of the data signals received within the second predefined time interval (e.g. in at least 50% + 1 of the data signals received within the second predefined time interval). In addition to the probabilistic algorithm, the module may preferably implement statistical techniques, such as mean or variance, to process the information produced by the probabilistic algorithm so as to generate the aforementioned output signal. Preferably, the module is configured to detect such anomaly in relation to a specific operating state of the electrical appliance. More preferably, the module is configured to generate an output signal indicative of the presence of said anomaly if the operating state represented in at least one of the received data signals differs from the operating state expected by the module. The term “module-predicted operating state” means the operating state predicted by the machine learning model or the operating state to which the probabilistic algorithm has assigned a probability greater than or equal to a pre-specified threshold. Preferably, the specific operating state of the electrical appliance may correspond to one of on, off, open, closed, active, and inactive. These features are particularly advantageous in order to optimize the operation of the module, in particular to optimize the detection of an anomaly in relation to specific operating states of the electrical appliance. In addition or as an alternative to the above features, the module is configured to generate an output signal indicating the absence of said anomaly if the data signals received correspond to those expected by the module. In at least one embodiment of the present invention, the electrical apparatus comprises at least one of: • at least one lighting device which preferably represents a first type of electrical appliance, • at least one shutter which preferably represents a second type of electrical appliance, • at least one contact device (preferably magnetic contact) to signal the opening or closing of a window or door which preferably represents a third type of electrical appliance, • at least one electrical socket which preferably represents a fourth type of electrical appliance, • at least one heating and, preferably, cooking device which preferably represents a fifth type of electrical appliance, • a thermal conditioning system designed to condition (directly or indirectly) the air temperature inside one or more rooms of a building which preferably represents a sixth type of electrical appliance, • at least one fan coil which preferably represents a seventh type of electrical appliance,• at least one thermostatic valve preferably representing an eighth type of electrical appliance, • at least one actuator for controlling valves of an air conditioning system preferably representing a ninth type of electrical appliance, • a device for measuring energy production or consumption preferably representing a tenth type of electrical appliance, and • a smart card reader device equipped with NFC / RFID preferably representing an eleventh type of electrical appliance. In at least one embodiment of the present invention, the electronic device is operatively connected to at least one lighting device and is configured to generate a data signal comprising data indicative of an operating state assumed by the at least one lighting device, in particular of a first operating state or a second operating state of the at least one lighting device. The lighting device is in the first operating state if it is on, i.e. if it is in operation to emit light (e.g. constant, dimmed or intermittent), while it is in the second operating state if it is off. In particular, if the electronic device is connected to a plurality of lighting devices then the electronic device is configured to generate a data signal comprising data indicative of a first operating state if at least one lighting device of the plurality of lighting devices is turned on, or a data signal comprising data indicative of a second operating state if all lighting devices of the plurality of lighting devices are turned off. In at least one embodiment of the present invention, the electronic device is operatively connected to at least one damper and is configured to generate a data signal comprising data indicative of an operating state assumed by the at least one damper, in particular of a first operating state or a second operating state of the at least one damper. A shutter is in the first operating state if it is open (partially or fully open) while it is in the second operating state if it is closed. Specifically, if the electronic device is connected to a plurality of shutters, the electronic device is configured to generate a data signal comprising data indicative of a first operating state if at least one shutter in the plurality of shutters is open, or a data signal comprising data indicative of a second operating state if all shutters in the plurality of shutters are closed. In at least one embodiment of the present invention, the electronic device is operatively connected to the thermal conditioning system and is configured to generate a data signal comprising data indicative of an operating state assumed by the thermal conditioning system, in particular a first operating state, a second operating state, a third operating state or a fourth operating state of the thermal conditioning system. The air conditioning system is in the first operating state if it is in operation and is set to heat the air temperature inside one or more rooms of a building; it is in the second operating state if it is in operation and is set to cool the air temperature inside one or more rooms of a building; it is in the third operating state if it is not in operation and is set to heat the air temperature inside one or more rooms of a building; and it is in the fourth operating state if it is not in operation and is set to cool the air temperature inside one or more rooms of a building. In at least one embodiment of the present invention, the electronic device comprises a thermostat arranged to detect an air temperature of a room. Additionally or alternatively, the thermostat can be set to detect the temperature of an element delimiting the room, such as a structural element of a room, in particular the room's screed. Preferably, the electronic device comprising the thermostat is operatively connected to the thermal conditioning system and is configured to generate a data signal comprising both data indicative of the operating state assumed by the thermal conditioning system and data indicative of the temperature detected by the thermostat. In at least one embodiment of the present invention, the system comprises a plurality of electronic devices operatively connected to the data processing device and to respective electrical appliances of the home automation system. In at least one embodiment of the present invention, electrical appliances connected to the plurality of electronic devices may assume at least a first operating state (e.g., on, open, or active) and a second operating state (e.g., off, closed, or inactive). Electrical appliances connected to a plurality of electronic devices define a set of electrical appliances. The set of electrical appliances can assume a plurality of predefined operating states. Preferably, the set of electrical appliances assumes a first operating state if at least one of the electrical appliances that form said set is in the first operating state, while it assumes a second operating state if all the electrical appliances of said set are in the second operating state. Preferably, the module is configured to detect an anomaly in the data signals received from the plurality of electronic devices and is configured to generate an output signal indicative of the presence of said anomaly based on said detection. This prediction therefore allows us to detect anomalies associated with a set of multiple electrical appliances. Preferably, this anomaly is detected by the module if the received data signals differ from those expected by the module. Preferably, this module is configured to detect said anomaly in relation to a specific operating state of the set of electrical appliances. Specifically, the module is configured to generate an output signal indicating the presence of said anomaly if the operating state of the set of electrical appliances differs from that expected by the module. In addition to or as an alternative to the plurality of electronic devices, the system preferably comprises at least one sensor device operatively connected to the data processing device via the communication network. The at least one sensor device comprises a sensor and is configured to transmit data signals over the communication network containing data indicative of a physical quantity detected by the sensor. Preferably, the sensor is included in the home automation system. In at least one embodiment of the present invention, the system comprises a plurality of sensor devices. This allows you to use the information provided by one or more sensors to detect anomalies. In at least one embodiment of the present invention, the sensor device comprises a sensor for detecting objects in an environment, such sensor device being configured to generate data signals comprising data indicative of the absence or presence of objects in such environment. In at least one embodiment of the present invention, the sensor device comprises a sensor for the motion of objects in an environment, such sensor device being configured to generate data signals comprising data indicative of motion in that environment. In at least one embodiment of the present invention, the sensor device comprises an air temperature sensor, such sensor device being configured to generate a data signal comprising data indicative of the air temperature detected in an environment. In at least one embodiment of the present invention, the data processing device comprises a plurality of modules. Preferably, each of said modules includes a related machine learning model or probabilistic algorithm. Each module of the plurality of modules is configured to detect an anomaly in the data signals received from a related group of devices consisting of at least one electronic device and / or at least one sensor device, and is configured to generate a related output signal indicative of the presence of said anomaly on the basis of said detection. The prediction of a plurality of modules advantageously allows for the identification of different anomalies, each associated with a relative group of devices. Preferably, said anomaly is detected by said module if the data signals received by it differ from those expected by said module. Preferably, electrical appliances belonging to the same device group are installed in the same area of ​​a building, for example in the same room. Preferably, the electrical appliances belonging to the same device group are of the same type, for example, all lighting devices or all shutters. The fact that a group of devices consists of electrical appliances of the same type advantageously allows us to define a homogeneous group of devices and therefore to obtain a machine learning module specialized in detecting anomalies in relation to this homogeneous group of devices. Preferably, the plurality of groups of devices comprises at least a first group of devices and a second group of devices, the type of electrical appliances belonging to the first group being different from the type of electrical appliances belonging to the second group. In at least one embodiment of the present invention, at least two modules of the aforementioned plurality of modules are configured to detect an anomaly in the data signals received from the same group of devices but in relation to different operating states of the electrical appliances belonging to that group of devices. In other words, the plurality of modules may comprise a first module configured to detect an anomaly in the data signals received from a group of devices and in relation to a first operating state of the electrical devices pertaining to that group of devices, and a second module configured to detect an anomaly in the data signals received from the same group of devices and in relation to a second operating state of the aforementioned electrical devices. By way of example, the first module may be configured to detect an anomaly in relation to an on state of lighting devices belonging to a first group of devices while the second module may be configured to detect an anomaly in relation to an off state of the aforementioned lighting devices. In at least one embodiment of the present invention, the data processing device is physically distinct from the electronic device of the system, or from each electronic device if the system comprises a plurality of electronic devices. Preferably, the data processing device is also physically distinct from said at least one sensor device if the system includes the latter. This feature advantageously allows the use of electronic devices (in particular control devices for electrical appliances) and sensor devices with low computing capacity, which are therefore economical to produce, since the detection of anomalies is entrusted to a separate entity, i.e. the data processing device. In at least one embodiment of the present invention, the home automation system includes the data processing device. Therefore, in this case, the data processing device is preferably installed in the building equipped with the home automation system. Alternatively, in at least one embodiment of the present invention, the data processing device is included in a cloud computing platform. Therefore, in this case, the data processing device is remote from the home automation system and the electrical device(s) included in the system to detect anomalies. This feature is particularly advantageous in terms of reducing the hardware capacity of electronic devices and / or electrical appliances. Furthermore, the provision of a cloud computing platform can provide advantages in terms of scalability, flexibility, and / or applicability to the electrical appliances present in the home automation system. Preferably, the system includes a gateway device configured to connect the home automation system's communication network to the network used by the cloud computing platform. In at least one embodiment of the present invention, the communication network of the home automation system comprises at least one of a wired network, a wireless network, a Bluetooth mesh network and a KNX network. In at least one embodiment of the present invention, the system, in particular the data processing device, is configured to generate an alarm signal if the output signal generated by the data processing unit module (or by at least one module if the data processing unit comprises a plurality of modules) indicates the presence of an anomaly in the data signals received therefrom. Preferably, the alarm signal corresponds to one of an audible signal, a visual signal, and a message transmitted via wireless communication, or a combination thereof. In at least one embodiment of the present invention, the system includes an interface intended for use by a user of the home automation system. The interface is operationally connected to the data processing device and configured to receive the aforementioned alarm signal. Preferably, the interface is included in a mobile device. With particular reference to the method for detecting anomalies, this method preferably includes: • operationally connect a plurality of electronic devices to the data processing device and to respective electrical appliances of the home automation system, and / or operationally connect at least one sensor device to the data processing device, the at least one sensor device comprising a sensor and being configured to transmit data signals in the communication network comprising data indicative of a physical quantity detected by said sensor, • define a plurality of groups of devices each consisting of at least one electronic device and / or at least one sensor device, wherein the data processing device comprises a plurality of modules, each module being configured to detect an anomaly in the data signals received from a relevant group of devices of the plurality of groups of devices, • transmit data signals in the communication network by means of the plurality of groups of devices,• receive in the modules the data signals transmitted by the respective groups of electronic devices, • detect by means of the modules whether an anomaly is present in the received data signals, • generate by means of the modules respective output signals indicating the presence of respective anomalies on the basis of the detections. Preferably, an anomaly is detected by one of the plurality of modules if the data signals received by that module differ from those expected by it. In at least one embodiment of the present invention, the method comprises retraining at least one machine learning model or re-determining the probabilities of the electrical appliance assuming an operational state at respective points in time based on data signals transmitted by a relevant group of electronic devices within a first predetermined time interval. Preferably, the at least one machine learning model is retrained at a first predetermined frequency. In at least one embodiment of the present invention, the method comprises evaluating the machine learning model at a predetermined execution frequency preferably greater than said first frequency. This is particularly advantageous for correctly detecting anomalies in particular periods of time, such as, for example, periods when a system user is away from the building incorporating the home automation system for a prolonged period. Brief description of the drawings The features and advantages of the present invention will be better understood from the detailed description of its preferred embodiments, illustrated by way of example and not by way of limitation with reference to the attached drawings, in which: • Figure 1 is a schematic view of a system for detecting anomalies according to an embodiment of the invention, • Figure 2 is a schematic view of a further system for detecting anomalies according to an embodiment of the invention, the system comprising a plurality of electronic devices and electrical appliances arranged in two distinct rooms of a building, and • Figure 3 is a flowchart representing a method for detecting anomalies according to an embodiment of the invention. Description of embodiments of the invention With reference to the attached figures, 100 or 100' generally indicates a system for detecting anomalies (hereinafter also referred to as system for brevity). With particular reference to Figure 1, the system 100 preferably comprises a home automation system 1 comprising a communication network 2 and an electrical appliance 3. The electrical appliance 3 is configured to assume a plurality of predefined operating states. The system 100 also comprises an electronic device 4 operatively connected to the communication network 2 and to the electrical appliance 3. The electronic device 4 is configured to transmit data signals S into the communication network 2 comprising data indicative of an operating state assumed by the electrical appliance 3 among the plurality of predefined operating states. In detail, the electronic device 4 comprises a control device 5 configured to control the operation of the electrical appliance 3 via the communication network 2. The system 100 further comprises a data processing device 6 operatively connected to the electronic device 4 via the communication network 2 and configured to receive data signals S transmitted by the electronic device 4. The data processing device 6 comprises a module, in particular corresponding to a machine learning model 7, the machine learning model 7 being trained to detect an anomaly in the data signals S received by the electronic device 4, in particular within a predetermined time interval, for example equal to at least 10 minutes. The machine learning model 7 (hereinafter also referred to as model for brevity) is also configured to generate an output signal O indicative of the presence of said anomaly on the basis of said detection. Figure 2 schematically shows a further example of a system for detecting anomalies, denoted by 100'. System 100' differs from System 100 described above in that it comprises a plurality of electronic devices, electrical appliances, and machine learning models. The plurality of electronic devices and electrical appliances are arranged in two distinct rooms A, B of a building E. In detail, in room A there is a first electrical appliance, i.e. a first lighting device 3a, a second electrical appliance, i.e. a first shutter 3b, and a third electrical appliance, i.e. a second shutter 3c. A first electronic device 4a, a second electronic device 4b, a third electronic device 4c and a fourth electronic device 4d are arranged in the environment A. The first electronic device 4a is operatively connected to the lighting device 3a and is configured to generate a data signal S1 comprising data indicative of a first operating state or a second operating state assumed by the first lighting device 3a, wherein the first lighting device 3a is in the first operating state if it is on and in the second operating state if it is off. Preferably, the data signal S1 also comprises data representing the time point at which the first electrical appliance is in the first operating state or in the second operating state. The second electronic device 4b is operationally connected to the first shutter 3b and is configured to generate a data signal S2 comprising data indicative of a first operating state or a second operating state assumed by the first shutter 3b, wherein the first shutter 3b is in the first operating state if it is open and in the second operating state if it is closed. Preferably, the data signal S2 also comprises data representing the time point in time at which the second electrical device is in the first operating state or in the second operating state. The third electronic device 4c is operationally connected to the second shutter 3c and is configured to generate a data signal S3 comprising data indicative of a first operating state or a second operating state assumed by the second shutter 3c, wherein the second shutter 3c is in the first operating state if it is open and in the second operating state if it is closed. Preferably, the data signal S3 also comprises data representing the time point in which the third electrical device is in the first operating state or in the second operating state. The first shutter 3b and the second shutter 3c form a first set of electrical devices. The first set of electrical devices assumes a first operating state if at least one of the first shutter 3b and the second shutter 3c is open, and a second operating state if the first shutter 3b and the second shutter 3c are closed. The electronic devices 4a, 4b and 4c include respective control devices 5 configured to control the operation of the respective electrical appliances 3a, 3b and 3c via the communication network 2. The fourth electronic device 4d comprises a thermostat T1 designed to detect the air temperature in room A and is operationally connected to a thermal conditioning system 3d, the latter representing a fourth electrical appliance of system 100'. The thermal conditioning system 3d is designed to condition the air temperature inside room A of building E. The 3D thermal conditioning system can assume a first operating state, a second operating state, a third operating state or a fourth operating state. The 3d thermal conditioning system is in the first operating state if it is in operation and is set to heat the air temperature inside room A, it is in the second operating state if it is in operation and is set to cool the air temperature inside room A, it is in the third operating state if it is not in operation and is set to heat the air temperature inside room A, while it is in the fourth operating state if it is not in operation and is set to cool the air temperature inside room A. The fourth electronic device 4d is also configured to generate a data signal S4 comprising both data indicative of the operating state assumed by the thermal conditioning system 3d and data indicative of the temperature detected by the thermostat T1. Preferably, the data signal S4 also comprises data representing the moment in time in which the fourth electrical appliance is in the first operating state, in the second operating state, in the third operating state, or in the fifth operating state, and data representing the moment in time in which the thermostat T1 detects the temperature. As for room B, it contains a fifth electrical appliance 3e1 + 3e2, consisting of a second lighting device 3e1 and a third lighting device 3e2, a sixth electrical appliance, i.e. a third shutter 3f, and a seventh electrical appliance, i.e. a fourth shutter 3g. A fifth electronic device 4e, a sixth electronic device 4f, a seventh electronic device 4g and an eighth electronic device 4h are arranged in room B. The fifth electronic device 4e is operatively connected to the second lighting device 3e1 and the third lighting device 3e2 and is configured to generate a data signal S5 comprising data indicative of a first operating state or a second operating state assumed by the fifth electrical appliance 3e1+3e2, wherein the fifth electrical appliance is in the first operating state if at least one of the second lighting device 3e1 and the third lighting device 3e2 is turned on, while it is in the second operating state if both lighting devices 3e1 and 3e2 are turned off. Preferably, the data signal S5 also comprises data representing the time point at which the fifth electrical appliance is in the first operating state or in the second operating state. The sixth electronic device 4f is operationally connected to the third shutter 3f and is configured to generate a data signal S6 comprising data indicative of a first operating state or a second operating state assumed by the third shutter 3f, wherein the third shutter 3f is in the first operating state if it is open and in the second operating state if it is closed. Preferably, the data signal S6 also comprises data representing the time point at which the sixth electrical device is in the first operating state or in the second operating state. The seventh electronic device 4g is operationally connected to the fourth shutter 3g and is configured to generate a data signal S7 comprising data indicative of a first operating state or a second operating state assumed by the fourth shutter 3g, wherein the fourth shutter 3g is in the first operating state if it is open and in the second operating state if it is closed. Preferably, the data signal S7 also comprises data representing the time point at which the seventh electrical device is in the first operating state or in the second operating state. The third damper 3f and the fourth damper 3g form a second set of electrical devices. The second set of electrical devices assumes a first operating state if at least one of the third damper 3f and the fourth damper 3g is open, and a second operating state if the third damper 3f and the fourth damper 3g are closed. The electronic devices 4e, 4f, and 4g include respective control devices 5 configured to control the operation of the respective electrical appliances 3e1 + 3e2, 3f, and 3g via the communication network 2. The eighth electronic device 4h comprises a thermostat T2 designed to detect the air temperature in room B and is configured to generate a data signal S8 containing data indicative of the temperature detected by thermostat T2. Preferably, the data signal S8 also includes data representing the moment in time at which thermostat T2 detects the temperature. System 100' also includes a sensor device DS1, which includes an object detection sensor. Sensor device DS1 is located in room B and is configured to generate a data signal S9, which includes data indicative of the absence or presence of objects in room B. Preferably, the system 100' comprises seven groups of devices defined as follows: a first group of devices corresponding to the first electronic device 4a, a second group of devices which is formed by the second electronic device 4b and the third electronic device 4c, a third group of devices corresponding to the fourth electronic device 4d, a fourth group of devices corresponding to the fifth electronic device 4e, a fifth group of devices which is formed by the sixth electronic device 4f and the seventh electronic device 4g, a sixth group of devices corresponding to the eighth electronic device 4h, and a seventh group of devices corresponding to the sensor device DS1. The data processing device 6 of the system 100' comprises a plurality of modules corresponding to respective machine learning models trained to detect an anomaly in the data signals received from respective groups of devices. In particular, the 100' system includes: • a first model, called first virtual sensor 7a, trained to detect an anomaly in the data signals S1 received from the first group of devices, in relation to the first operating state of the first lighting device 3a, • a second model, called second virtual sensor 7b, trained to detect an anomaly in the data signals S1 received from the first group of devices, in relation to the second operating state of the first lighting device 3a, • a third model, called third virtual sensor 7c, trained to detect an anomaly in the data signals S2 and S3 received from the second group of devices, in relation to the first operating state of the first set of electrical appliances, • a fourth model, called fourth virtual sensor 7d, trained to detect an anomaly in the data signals S2 and S3 received from the second group of devices, in relation to the second operating state of the first set of electrical appliances, • a fifth model,called fifth virtual sensor 7e, trained to detect an anomaly in the data signals S4 received from the third group of devices, in relation to the first operating state of the thermal conditioning system 3d and the temperature (in particular the temperature trend) detected by the thermostat T1, • a sixth model, called sixth virtual sensor 7f, trained to detect an anomaly in the data signals S4 received from the third group of devices, in relation to the second operating state of the thermal conditioning system 3d and the temperature (in particular the temperature trend) detected by the thermostat T1, • a seventh model, called seventh virtual sensor 7g, trained to detect an anomaly in the data signals S4 received from the third group of devices, in relation to the third operating state of the thermal conditioning system 3d and the temperature (in particular the temperature trend) detected by the thermostat T1, • an eighth model,called eighth virtual sensor 7h, trained to detect an anomaly in the data signals S4 received from the third group of devices, in relation to the fourth operating state of the thermal conditioning system 3d and to the temperature (in particular the temperature trend) detected by the thermostat T1, • a ninth model, called ninth virtual sensor 7i, trained to detect an anomaly in the signals S5 received from the fourth group of devices, in relation to the first operating state of the fifth electrical appliance 3e1+3e2, • a tenth model, called tenth virtual sensor 7l, trained to detect an anomaly in the signals S5 received from the fourth group of devices, in relation to the second operating state of the fifth electrical appliance 3e1+3e2, • an eleventh model, called eleventh virtual sensor 7m, trained to detect an anomaly in the data signals S6 and S7 received from the fifth group of devices,in relation to the first operating state of the second set of electrical appliances, • a twelfth model, called the twelfth virtual sensor 7n, trained to detect an anomaly in the data signals S6 and S7 received from the fifth group of devices, in relation to the second operating state of the second set of electrical appliances, • a thirteenth model, called the thirteenth virtual sensor 7o, trained to detect an anomaly in the data signals S8 received from the sixth group of devices, in relation to the temperature detected by the thermostat T2, • a fourteenth model, called the fourteenth virtual sensor 7p, trained to detect an anomaly in the signals S9 received from the seventh group of devices, in relation to the absence of objects in the environment B, and • a fifteenth model, called the fifteenth virtual sensor 7q, trained to detect an anomaly in the signals S9 received from the seventh group of devices,in relation to the presence of objects in environment B., A further model can also be envisaged, i.e. a further virtual sensor, trained to detect an anomaly in the data signals received from thermostat T1 and indicative of the temperature it detects. Furthermore, each model of the plurality of machine learning models 7a-7q is configured to generate a corresponding output signal indicative of the presence of said anomaly if the latter is detected by such machine learning model. Specifically, the first model 7a is configured to generate a first output signal O1 indicative of the presence of an anomaly if the data signals received by it represent the second operating state of the first lighting device 3a while the operating state expected by the first virtual sensor 7a corresponds to the first operating state. The second model 7b is configured to generate a second output signal O2 indicative of the presence of an anomaly if the data signals received by it represent the first operating state of the first lighting device 3a while the operating state expected by the second virtual sensor 7b corresponds to the second operating state. The third model 7c is configured to generate a third output signal O3 indicative of the presence of an anomaly if the data signals received by it represent the second operating state of the first set of electrical devices while the operating state expected by the third virtual sensor 7c corresponds to the first operating state. The fourth model 7d is configured to generate a fourth output signal O4 indicative of the presence of an anomaly if the data signals received by it represent the first operating state of the first set of electrical devices while the operating state expected by the third virtual sensor 7d corresponds to the second operating state. The fifth model 7e is configured to generate a fifth output signal O5 indicating the presence of an anomaly if the data signals received by it represent the first operating state of the 3d thermal conditioning system and a failure to increase the temperature detected by thermostat T1 while the fifth virtual sensor 7e expects the 3d thermal conditioning system to be in the first operating state and that the temperature detected by thermostat T1 increases. The sixth model 7f is configured to generate a sixth output signal O6 indicating the presence of an anomaly if the data signals received by it represent the second operating state of the thermal conditioning system 3d and a failure to decrease the temperature detected by thermostat T1 while the sixth virtual sensor 7f expects the thermal conditioning system 3d to be in the second operating state and that the temperature detected by thermostat T1 decreases. The seventh model 7g is configured to generate a seventh output signal O7 indicating the presence of an anomaly if the data signals received by it represent the third operating state of the 3d thermal conditioning system and an increase in the temperature detected by thermostat T1 while the seventh virtual sensor 7g assumes that the 3d thermal conditioning system is in the third operating state and that the temperature detected by thermostat T1 does not increase. The eighth model 7h is configured to generate an eighth output signal O8 indicating the presence of an anomaly if the data signals received by it represent the fourth operating state of the 3d thermal conditioning system and a decrease in the temperature detected by thermostat T1 while the eighth virtual sensor 7h assumes that the 3d thermal conditioning system is in the fourth operating state and that the temperature detected by thermostat T1 does not decrease. The ninth model 7i is configured to generate a ninth output signal O9 indicating the presence of an anomaly if the data signals received by it represent the second operating state of the fifth electrical appliance 3e1+3e2 while the operating state expected by the ninth virtual sensor 7i corresponds to the first operating state. The tenth model 7l is configured to generate a tenth output signal O10 indicating the presence of an anomaly if the data signals received by it represent the first operating state of the fifth electrical appliance 3e1+3e2 while the operating state expected by the tenth virtual sensor 7l corresponds to the second operating state. The eleventh model 7m is configured to generate an eleventh output signal O11 indicating the presence of an anomaly if the data signals received by it represent the second operating state of the second set of electrical appliances while the operating state expected by the eleventh virtual sensor 7m corresponds to the first operating state. The twelfth model 7n is configured to generate a twelfth output signal O12 indicative of the presence of an anomaly if the data signals received by it represent the first operating state of the second set of electrical devices while the operating state expected by the twelfth virtual sensor 7n corresponds to the second operating state. The thirteenth model 7o is configured to generate a thirteenth output signal O13 indicating the presence of an anomaly if the temperature detected by thermostat T2 differs from the temperature expected by the thirteenth virtual sensor 7o. The fourteenth model 7p is configured to generate a fourteenth output signal O14 indicating the presence of an anomaly if the data signals received by it indicate the presence of objects in environment B while the fourteenth model 7p predicts the absence of objects in environment B. The fifteenth model 7q is configured to generate a fifteenth output signal O15 indicative of the presence of an anomaly if the data signals received by it indicate the absence of objects in environment B while the fifteenth model 7q predicts the presence of objects in environment B. The data processing device 6 of the system 100' is physically distinct from the electronic devices 4a-4h and the sensor device DS1 because it is included in a cloud computing platform 8. It should be noted that the virtual sensors thus obtained, advantageously residing in a cloud computing platform, can be used by further authorized data processing systems to perform specific analyses or tasks, for example based on the O1-15 output signals. Systems 100 and 100', illustrated in Figures 1 and 2, have been described in relation to one or more modules corresponding to respective machine learning models. However, said one or more modules may correspond to respective probabilistic algorithms in place of the machine learning models, said probabilistic algorithms falling within the scope of protection defined by the invention as claimed. An example of the evaluation of the probabilistic algorithm based on ten data signals received on February 6, 2024 is presented below. Table 1 shows as an example the operating states of the first lighting device 3a in relation to a period [0,9] equal to nine minutes extracted from February 6, 2024. In the period [0,9] the module comprising the probabilistic algorithm therefore receives ten data signals: the first data signal at minute 0 and the subsequent 9 data signals with intervals of one minute between them. Table 1 minute Operating state (1 = first operating state; 0 = second operating state) 0 1 1 1 2 1 3 0 4 0 5 1 6 0 7 0 8 0 9 0 Tables 2 and 3 below show, in illustrative terms, the probabilities that the first lighting device 3a assumes the first operating state (on) and the second operating state (off) over the same ten-minute period. Specifically, the probabilities of each column in the table were determined by the probabilistic algorithm 10 based on data signals received at different time intervals during the training phase, namely: general_tab column: probabilities determined by the probabilistic algorithm based on all the data sent as input to the module in the training phase, winter_tab column: probabilities determined by the probabilistic algorithm based on the data relating to the winter season and sent as input to the module in the training phase, February_tab column: probabilities determined by the probabilistic algorithm based on the data relating to the month of February and sent as input to the module in the training phase, Tuesday_tab column: probabilities determined by the probabilistic algorithm based on the data relating to Tuesdays and sent as input to the module in the training phase, weekday_tab column: probabilities determined by the probabilistic algorithm based on the data relating to weekdays and sent as input to the module in the training phase. The columns tab_inverno, tab_febbraio, tab_martedì and 15 tab_feriale were extracted by the probabilistic algorithm from the set of columns relating to the season, month, day of the week and type of day (weekday / holiday) as they are relevant to the day 6 February 2024 in question. Table 2 Probability of the first operational state (between 0 and 1) minute general_tab winter_tab February_tab Tuesday_tab weekday 0 0.001 0.002 0.09 0.0011 0.0009 1 0.0012 0.0021 0.11 0.0012 0.001 2 0.0011 0.002 0.1 0.0011 0.0012 3 0.001 0.002 0.12 0.001 0.001 4 0.0012 0.0022 0.15 0.009 0.001 5 0.95 0.94 0.98 0.99 0.901 6 0.001 0.002 0.09 0.0011 0.0009 7 0.0012 0.0021 0.11 0.0012 0.001 8 0.0011 0.002 0.1 0.0011 0.0012 9 0.001 0.002 0.12 0.001 0.001 Table 3 Probability of the second operating state (between 0 and 1) minute general_tab winter_tab February_tab Tuesday_tab weekday 0 0.999 0.998 0.91 0.9989 0.9991 1 0.9988 0.9979 0.89 0.9988 0.999 2 0.9989 0.998 0.9 0.9989 0.9988 3 0.999 0.998 0.88 0.999 0.999 4 0.9988 0.9978 0.85 0.991 0.999 5 0.05 0.06 0.02 0.01 0.099 6 0.999 0.998 0.91 0.9989 0.9991 7 0.9988 0.9979 0.89 0.9988 0.999 8 0.9989 0.998 0.9 0.9989 0.9988 9 0.999 0.998 0.88 0.999 0.999 Setting the pre-established threshold equal to 0.9, the following table instead shows the detection of anomalies by the probabilistic algorithm in relation to the operating states assumed by the first lighting device 3a in the period [0.9] and to the different time intervals described above, where 0 represents the presence of an anomaly while 1 represents the absence of an anomaly. Table 4 minute operating status Anomaly feedback from general_tab anomaly winter_tab anomaly February_tab anomaly Tuesday_tab anomaly weekday_tab 0 1 0 0 0 0 0 1 1 0 0 0 0 0 2 1 0 0 0 0 0 3 0 1 1 0 1 1 4 0 1 1 0 1 1 5 1 1 1 1 1 1 6 0 1 1 1 1 1 7 0 1 1 0 1 1 8 0 1 1 1 1 9 0 1 1 0 1 1 In particular, the minute 0 row shows the presence of an anomaly for each anomaly column because the probability of the first operational state in that minute was less than the threshold for each column in Table 2. Conversely, the minute 3 row shows the presence of an anomaly for only the anomaly column tab_febbraio because the probability of the second operational state in that minute was less than the threshold only for the tab_febbraio column in Table 3. The following table represents the conclusive detection of anomalies 10 by the probabilistic algorithm in relation to the operating states assumed by the first lighting device 3a in the period [0,9], where 0 in the anomaly column represents the presence of an anomaly while 1 represents the absence of an anomaly. In this case, an anomaly is detected in reference to a specific data signal if the majority of the anomaly columns in Table 4 have a 0 in relation to the same data signal, i.e. in relation to the same minute of the period [0,9]. Table 5 minute operating state anomaly 0 1 0 1 1 0 2 1 0 3 0 1 4 0 1 5 1 1 6 0 1 7 0 1 8 0 1 9 0 1 Finally, by way of example, the probabilistic algorithm is configured to generate the output signal indicating the presence of an anomaly within the period [0,9] of 6 February 2024 if the anomaly is detected by the probabilistic algorithm in at least a pre-established 10 percentage of the data signals received in the period [0,9] (for example in at least 50% + 1 of the data signals received). Figure 3 shows a flowchart representing a method 200 for detecting anomalies according to one embodiment of the invention. The method 200 comprises a step 201 of operatively connecting the electronic device 4 to the communication network 2 of a home automation system 1, a step 202 of operatively connecting the electronic device 4 to the electrical appliance 3, and a step 203 of operatively connecting the data processing device 6 to the electronic device 4 via the communication network 2. The method 200 further comprises a step 204 of transmitting data signals S into the communication network 2 by means of the electronic device 4, a step 205 of receiving in the module 7 of the data processing device 6 the data signals S transmitted by the electronic device 4 and detecting by means of this module whether an anomaly is present in the data signals S, and a step 206 of generating by means of the machine learning model 7 an output signal O indicative of the presence of said anomaly on the basis of said detection. With particular reference to figure 2, method 200 may also include the following phases: • operatively connecting a plurality of electronic devices 4a-4h to the data processing device 6 and respective electrical appliances 3a-3g, and operatively connecting the sensor device DS1 to the data processing device 6, • defining a plurality of device groups each consisting of at least one electronic device and / or the sensor device DS1, wherein said data processing device comprises a plurality of modules, in particular corresponding to respective machine learning models 7a-7q, each module of said plurality of modules being configured to detect an anomaly in the data signals received from a respective device group of the plurality of device groups, • transmitting data signals S1-S9 in the communication network 2 by means of the plurality of device groups,• receive in modules 7a-7q the data signals transmitted by the respective groups of electronic devices and evaluate said machine learning models on the basis of the received data signals, and • detect by means of said modules whether an anomaly is present in the received data signals, • generate by means of modules 7a-7q respective OO15 output signals indicative of the presence of respective anomalies on the basis of said detections., In the case of modules corresponding to machine learning models, it may also be envisaged to retrain at least one machine learning model on the basis of data signals transmitted by a relevant group of electronic devices over a given period of time. Preferably, at least one machine learning model is retrained every 7 days. Preferably, the machine learning model is evaluated every 10 minutes.

Claims

1. System (100;100') for detecting anomalies comprising: • at least one electronic device (4) operatively connected to a communication network (2) and to an electrical appliance (3) of a home automation system (1), said electrical appliance being configured to assume a plurality of predefined operating states, wherein said at least one electronic device is configured to transmit in said communication network data signals (S) comprising data indicative of an operating state assumed by the electrical appliance among said plurality of predefined operating states, and • a data processing device (6) operatively connected to said at least one electronic device through said communication network and configured to receive data signals transmitted by said at least one electronic device,wherein said data processing device comprises a module configured to detect an anomaly in the data signals (S) received from said at least one electronic device and to generate an output signal (O) indicative of the presence of said anomaly based on said detection., 2. System according to claim 1, wherein said module comprises one of: • a machine learning model (7) trained to detect said anomaly in the data signals (S) received from said at least one electronic device, and • a probabilistic algorithm configured to detect said anomaly in the data signals (S) received from said at least one electronic device based on at least one comparison of a probability that said electrical appliance assumes the operating state represented in said data signals (S) with a predetermined threshold.

3. System according to one of the preceding claims, wherein said at least one electronic device comprises at least one control device (5) configured to control the operation of said electrical appliance.

4. System according to one of the preceding claims, wherein said module is configured to detect said anomaly in relation to a specific operating state of said electrical appliance.

5. System according to one of the preceding claims, wherein said electrical appliance comprises at least one of: • at least one lighting device (3a, 3e1, 3e2), • at least one shutter (3b, 3c, 3f, 3g), • at least one contact device to signal the opening or closing of a window, • at least one electrical socket, • at least one heating and, preferably, cooking device, • a thermal conditioning system (3d) designed to condition the air temperature inside one or more rooms of a building, • at least one fan coil, • at least one thermostatic valve, • at least one actuator for controlling valves of an air conditioning system, • a device for measuring energy production or consumption, and • a smart card reader device equipped with NFC / RFID.

6. System according to one of the preceding claims, wherein said system comprises a plurality of electronic devices (4a,..., 4h) operatively connected to said data processing device and to respective electrical appliances (3a,..., 3g) of said home automation system, said module being configured to detect an anomaly in the data signals received from said plurality of electronic devices and to generate an output signal indicative of the presence of said anomaly on the basis of said detection.

7. System according to one of the preceding claims, wherein said system comprises at least one sensor device (DS1) operatively connected to said processing device via said communication network, said at least one sensor device comprising a sensor and being configured to transmit data signals in said communication network comprising data indicative of a physical quantity detected by said sensor.

8. The system of claim 7, wherein said sensor is a sensor for detecting objects in an environment, said sensor device (DS1) being configured to generate data signals comprising data indicative of the absence or presence of objects in said environment.

9. System according to one of the preceding claims, wherein said data processing device comprises a plurality of modules, each module being configured to detect an anomaly in the data signals (S1, ..., S9) received from a related group of devices consisting of at least one electronic device and / or at least one sensor device, and being configured to generate a related output signal (O1, ..., O15) indicative of the presence of said anomaly on the basis of said detection.

10. System according to one of the preceding claims, wherein said data processing device is physically distinct from each electronic device and from said at least one sensor device if said system comprises the latter.

11. System according to one of the preceding claims, wherein said data processing device is included in a cloud computing platform (8).

12. A method for detecting anomalies, the method (200) comprising: • operatively connecting an electronic device (4) to a communication network (2) of a home automation system (1), • operatively connecting said electronic device to an electrical appliance (3) of said home automation system, said electrical appliance being configured to assume a plurality of predefined operating states and said at least one electronic device being configured to transmit into the communication network data signals (S) comprising data indicative of an operating state assumed by the electrical appliance among said plurality of predefined operating states, • operatively connecting a data processing device (6) to said electronic device via said communication network, wherein said data processing device comprises a module, said module being configured to detect an anomaly in the data signals received from said electronic device,• transmit data signals in said communication network by means of said electronic device, • receive in said module the data signals transmitted by said electronic device, • detect by means of said module whether an anomaly is present in said received data signals, • generate by means of said module an output signal (O) indicative of the presence of said anomaly on the basis of said detection., 13. Method according to claim 12, comprising: • operatively connecting a plurality of electronic devices (4a,..., 4h) to said data processing device and to respective electrical appliances (3a,..., 3g) of said home automation system, and / or operatively connecting at least one sensor device (DS1) to said data processing device, said at least one sensor device comprising a sensor and being configured to transmit data signals in said communication network comprising data indicative of a physical quantity detected by said sensor, • defining a plurality of groups of devices each formed by at least one electronic device and / or by at least one sensor device, wherein said data processing device comprises a plurality of modules, each module being configured to detect an anomaly in the data signals (S1, .„, S9) received from a respective group of devices of said plurality of groups of devices, • transmit data signals in said communication network by means of said plurality of groups of devices, • receive in said modules the data signals transmitted by the respective groups of electronic devices, • detect by means of said modules whether an anomaly is present in said received data signals, • generate by means of said modules respective output signals (O1, ..., O15) indicative of the presence of respective anomalies on the basis of said detections.