System and method for controlling snow making machines

Abstract

The present invention relates to a system and a method for controlling snow making machines (40) on a ski slope (44) and involves a system for predicting snow depth on a ski slope (44) comprising a data receiver module; a machine learning module; an application module; a prediction module; and a control module. The data receiver module is configured to receive historical snow data related to the ski slope, including recorded measurements of snow depth at various locations on the ski slope and associated ambient conditions at the time of each measurement. Furthermore, the data receiver module is configured to receive current snow data and current ambient condition data for the ski slope. The machine learning module is configured to train a machine learning algorithm with the historical snow data and to generate a prediction model for forecasting snow depth evolution on the ski slope (44). The application module is configured to apply the current snow data and the current ambient condition data to the prediction model. The prediction module is configured to predict snow depth on the ski slope (44) using the prediction model. The control module is configured to generate, based on the predicted snow depth, at least one of a) control signals for automatic transmission to the snow making machines (40) to control at least one of starting snow production, stopping snow production, adjusting snow goals like specified snow amounts, adjusting water flow rates, and adjusting air pressure, and control information for presentation on a user interface for manual control of the snow making machines (40).

Description

[0001] M / TAL-012-PC

[0002] - 1 -

[0003] SYSTEM AND METHOD FOR CONTROLLING SNOW MAKING MACHINES

[0004] TECHNICAL FIELD

[0005] The present invention relates to technologies for snow management, and more particularly a system and a method for controlling snow making machines employing a machine learning.

[0006] TECHNICAL BACKGROUND

[0007] Artificial snowmaking systems are commonly used in ski resorts to ensure adequate snow coverage on ski slopes, particularly during periods of insufficient natural snowfall. These systems typically involve the use of snowmaking machines, also known as "snow cannons", which generate artificial snow by spraying a mixture of water and compressed air into the cold environmental air. The water droplets freeze in the air and fall to the ground as snow, thereby increasing the snow depth on the ski slopes.

[0008] The operation of these snowmaking systems is influenced by various factors, including the ambient temperature, relative humidity, wind speed and direction, and, if the snowmaking is done during daytime, which may happen in particular in pre-season preparation, the energy given by the incident sunlight. These factors can affect the quality and quantity of the artificial snow produced, as well as the energy efficiency of the snowmaking process.

[0009] In order to manage the operation of the snowmaking systems, ski resorts typically rely on manual decision-making processes based on the operators' experience and judgment. For instance, the decision to start or stop the snowmaking process, or to adjust the settings of the snowmaking machines, is often made by the operators based on their assessment of the current snow conditions and weather forecasts. However, this manual approach to snow management can be challenging and inefficient, as it requires constant monitoring of the snow conditions and weather forecasts, and it may not accurately predict the future snow depth on the ski slopes. Furthermore, it does not take into account the historical snow data and slope- M / TAL-012-PC

[0010] - 2 - specific factors, such as sun exposure, rain exposure, fog, slope gradient, base composition, skier movement and skier frequency, which can also influence the snow depth evolution.

[0011] Existing snowmaking systems, such as those described in US 4,7171,072 A, involve a network of snow making machines along a ski run, each fed with snowmaking liquid and controlled by a central processing unit. While this document outlines a general snow production system, it lacks predictive capabilities regarding future snow conditions. Decisions on when and how much snow to produce are made manually, relying on operator judgment and experience.

[0012] A very beneficial and useful enhancement to this approach is presented in EP 2 713 119 B2, which introduces a feedback system that utilizes real-time data to refine the control of snowmaking machines. This system can integrate weather forecasts to inform the user, via a graphical interface, about upcoming weather conditions.

[0013] However, the control of snowmaking operations still remains a manual process, with decisions based on the user's interpretation of the forecasts.

[0014] These traditional methods of snow management are limited by their reactive nature and reliance on human decision-making. They do not account for the dynamic evolution of snow depth over time or the complex interplay of changing weather conditions. Consequently, there is a need for a more proactive, precise, and automated approach to snow management.

[0015] Revilloud, Marc et al.: "Predicting snow height in ski resorts using an agent-based simulation" in Multiagent and Grid Systems, Vol. 9, No. 4, 10 January 2014, pages 279 - 283 discloses a system for predicting snow height on ski runs through agentbased simulation. The approach combines interpolated snow height measurements obtained using a neural network, local meteorological forecasts for every ski resort, and a simulation of skier activity in order to estimate the evolution of snow cover.

[0016] CN 116879979 A (Jiangsu Meteorological Observatory) discloses a method and system for predicting snow depth using meteorological factor vectors, a factor attention mechanism, and a spatiotemporal neural network that captures both M / TAL-012-PC

[0017] - 3 - spatial and temporal dependencies. The system includes preprocessing and forecasting modules and can be implemented on a computer device or storage medium to output predicted snow accumulation efficiency and snow depth for meteorological forecasting and disaster prevention.

[0018] BRIEF SUMMARY OF THE INVENTION

[0019] Despite advancements in snow management systems, existing approaches exhibit several technical shortcomings. Traditional systems depend on real-time data and weather forecasts to guide snow production, yet they lack automated control over snow making machines. This deficiency forces users to rely on their judgment and experience for decision-making. Such systems do not provide a comprehensive understanding of snow depth evolution, failing to consider historical snow data, slope-specific factors like sun exposure, slope angle, and skier-related impacts, nor do they account for forecasts of dynamic factors like skier numbers that can substantially affect snow depth. Consequently, these limitations can lead to inefficient snow management, resulting in the unnecessary expenditure of resources and energy.

[0020] The invention addresses the technical problem of providing a more holistic approach that does not rely on human experience and that can accurately predict snow depth on ski slopes using machine learning, taking into account both historical data and various forecasts, and can automate the control of snow making machines based on these predictions. The invention also addresses the urgent need for enhancing environmental sustainability by minimizing waste and optimizing the use of water and energy in snowmaking processes. Additionally, it aims at conserving resources by accurately predicting snow depth and distribution requirements, thereby reducing the environmental impact and operational costs associated with snow production.

[0021] The technical problems are solved by a predictive system for controlling snow making machines as described in claim 1, and a method for controlling snow making machines according to claim 11. The respective dependent claims detail advantageous embodiments and enhancements. M / TAL-012-PC

[0022] - 4 -

[0023] The invention provides multiple technical advantages. It enables better control of the amount of snow made and the timing of snow making, which can result in substantial energy and resource savings, allowing for the production of as little artificial snow as is strictly necessary.

[0024] A particular advantage of the invention is that it can be used for retrofitting existing snow making machines and snow management systems, thereby enhancing their functionality without the cost and inconvenience of replacing snow making machines and other hardware. This makes the invention a cost-effective solution for ski resorts looking to improve their snow management practices. Obviously, the invention cannot only be used to control individual or groups of individual snowmaking machines in a fully or semi-automated manner, but also to control entire snowmaking plants of ski resorts, i.e. a multitude of different snowmaking machines distributed over a large area having a multitude of different slope-specific requirements for snow production.

[0025] In contrast to for example the approach provided in the document by Revilloud et al. cited above, which is a hybrid model (GRNN interpolation with meteorological and agent-based skier simulation), which is formally a type of statistical model from the neural network family, the invention is trained with large historical datasets to recognize patterns. In Revilloud, the GRNN is only a tool for spatial smoothing (interpolation), especially when measurement points are missing, whereas the invention is a learning system that is dynamically / continuously adapted and optimized during operation. Therefore, Revilloud only provides forecasts, but no integration into operational control of snowmaking systems. The inventive approach, in contrast, is a data-driven machine learning model, preferably with realtime feedback loop and automatic control of snowmaking systems. No physical model or agent simulation as in Revilloud is used, but rather a statistical prediction model that learns patterns from historical and real-time data.

[0026] The invention also provides a holistic perspective on snow depth evolution, not limited to the current state but extending to future developments. This predictive foresight, which allows the application of the model to multiple ski slopes with M / TAL-012-PC

[0027] - 5 - different characteristics, enables proactive planning and decision-making, resulting in more efficient resource utilization and improved slope conditions.

[0028] Furthermore, the invention can be adapted for a variety of applications beyond ski resorts. For instance, it can be used to manage snow conditions for winter sports events where maintaining consistent and safe snow conditions is paramount for competition. Snow parks can benefit from the technology to precisely control snow depth for features such as halfpipes and jumps. For mountain roads, accurate snow depth predictions are invaluable for road safety and maintenance planning. Airports in cold climates can use the predictions to better plan snow removal operations, minimizing delays and ensuring the safety of aircraft operations. Construction sites operating during winter months can predict and manage snow accumulation, reducing work stoppages and improving safety. The invention's versatility extends to any scenario where snow depth can impact operations, safety, or resource allocation, making it a valuable tool across various industries. The invention's adaptability also means it can be customized to suit the specific requirements of different environments and applications, providing tailored solutions for diverse snow management challenges.

[0029] The present invention encompasses a system for controlling snow making machines on a ski slope, which includes several interconnected components designed to process and analyze snow-related data for effective snow management. The system comprises a data receiver module that is tasked with collecting historical snow data pertinent to the ski slope. This data encompasses a record of snow depth measurements taken at various locations on the ski slope, along with the ambient conditions present at the time of each measurement.

[0030] A machine learning module forms a core part of the system, equipped to train a machine learning algorithm using the historical snow data. The trained algorithm is then utilized to create a prediction model capable of forecasting the evolution of snow depth on the ski slope. To ensure the model remains current and accurate, the data receiver module is also configured to acquire up-to-date snow data and information regarding the current ambient conditions affecting the ski slope. M / TAL-012-PC

[0031] - 6 -

[0032] An application module is responsible for integrating the current snow data and ambient condition data into the prediction model, allowing for real-time updates to the forecasts. The prediction module uses this refined prediction model to ascertain future snow depth across the ski slope.

[0033] Finally, an output module is included to convey information and recommendations derived from the predicted snow depth. These outputs are designed to guide snowmaking and snow grooming operators in making informed decisions regarding snow production and maintenance, ultimately enhancing the efficiency and safety of ski slope operations.

[0034] This system offers the technical advantage of providing a proactive approach to snow management, leveraging historical and current data to inform snow depth predictions. The use of a machine learning algorithm allows for the analysis of complex patterns in snow depth changes, leading to more accurate forecasts that can guide efficient snowmaking and maintenance decisions. By outputting actionable recommendations, the method facilitates informed decision-making that can enhance the safety and enjoyment of ski slope users while optimizing resource utilization.

[0035] According to one embodiment, the system for controlling snow making machines on a ski slope comprises a system for predicting snow depth on a ski slope comprising a data receiver module; a machine learning module; an application module; a prediction module; and a control module. The data receiver module is configured to receive historical snow data related to the ski slope, including recorded measurements of snow depth at various locations on the ski slope and associated ambient conditions at the time of each measurement. Furthermore, the data receiver module is configured to receive current snow data and current ambient condition data for the ski slope. The machine learning module is configured to train a machine learning algorithm with the historical snow data and to generate a prediction model for forecasting snow depth evolution on the ski slope. The application module is configured to apply the current snow data and the current ambient condition data to the prediction model. The prediction module is configured to predict snow depth on the ski slope using the prediction model. The control M / TAL-012-PC

[0036] - 7 - module is configured to generate, based on the predicted snow depth, at least one of a) control signals for automatic transmission to the snow making machines to control at least one of starting snow production, stopping snow production, adjusting snow goals like specified snow amounts, adjusting water flow rates, and adjusting air pressure, and control information for presentation on a user interface for manual control of the snow making machines.

[0037] This system has the technical advantage of enabling precise and proactive snow management by utilizing historical and current data to inform snow depth predictions. The use of a machine learning algorithm allows for the analysis of complex patterns in snow depth changes, leading to more accurate forecasts that can guide efficient snowmaking and maintenance decisions. The snowmaking machines may be automatically controlled without requiring manual intervention or alternatively may be manually operated. By outputting actionable recommendations, the method facilitates informed decision-making that can enhance the safety and enjoyment of ski slope users while optimizing resource utilization.

[0038] As used herein, "historical snow data" denotes recorded measurements of snow depth and ambient conditions from past seasons, which provide a reference for understanding snow behavior under various conditions. "Machine learning algorithm" refers to a computational model capable of learning from data to identify patterns and make predictions. "Prediction model" denotes the outcome of the machine learning algorithm, which is a tool for estimating future snow depth based on learned data. "Current snow data" includes recent measurements of snow depth and ambient conditions, which are used to update and refine the prediction model.

[0039] The training of the machine learning algorithm involves feeding it with historical snow data, which includes measurements of snow depth and ambient conditions such as for example temperature, wind speed, and humidity from past seasons. This data acts as a training set, where the algorithm learns to recognize patterns and correlations between the various factors and the resulting snow depth. The training process may include dividing the data into a training set and a validation set, using the former to build the model and the latter to test its accuracy. Techniques such as cross-validation may be employed to ensure that the model generalizes well to new, M / TAL-012-PC

[0040] - 8 - unseen data. Over time, the algorithm iteratively adjusts its internal parameters to minimize prediction errors, improving its ability to forecast snow depth accurately.

[0041] Several types of machine learning algorithms can be used for predicting snow depth, each with its own strengths and suitable applications. These may include neural networks, which are particularly good at capturing complex, non-linear relationships in large datasets; decision trees, which provide clear, rule-based predictions; support vector machines, which are effective in high-dimensional spaces; and random forest algorithms, which combine multiple decision trees to improve prediction accuracy and control over-fitting. The choice of algorithm depends on the specific characteristics of the data and the desired outcomes of the prediction model.

[0042] A prediction model is the result of the machine learning algorithm's training process. It is a mathematical representation of the relationships learned from the historical snow data. The model takes input variables, such as current ambient conditions and snow depth measurements, and outputs a prediction of future snow depth. The complexity of the model can vary, from simple linear regressions to complex ensembles of models. The prediction model is used to estimate the snow depth at various locations on the ski slope over time, providing a forecast that can inform snow management decisions.

[0043] Applying current snow data to the prediction model involves updating the model with the latest measurements of snow depth and ambient conditions. This real-time data is used to refine the model's predictions, making them more accurate and relevant to the current state of the ski slope. The application of current data can be done in various ways, such as retraining the model with a combination of historical and new data or using techniques like online learning, where the model is continuously updated as new data comes in. This ensures that the prediction model remains up-to-date and can provide the ski slope operators with the latest information for proactive snow management.

[0044] In an embodiment of the system for controlling snow making machines, the data receiver module is further configured to receive real-time feedback regarding actual M / TAL-012-PC

[0045] - 9 - snow data. "Snow data" refers to data of snow made freshly by the snow making machines and / or data of natural snow. In addition, the machine learning module is configured to continuously update the prediction model based on the real-time feedback and / or the prediction module is configured to continuously update the prediction model based on the real-time feedback.

[0046] The feedback may include data on the efficacy of snow production by the snow making machines, thereby refining the prediction model for increased accuracy in subsequent forecasts. As will be understood by a person skilled in the art, in practice sensor data is often averaged over a certain period of time, ranging for example from several seconds to several minutes, to prevent "over-controlling" and adjusting the machines to often. Hence, as used herein and as customary in the art, the term real-time feedback is used to denote actual immediate feedback as well as feedback based on data averaged over a certain short period of time.

[0047] The technical advantage of this aspect is the enhanced accuracy of the snow depth predictions, which is achieved by incorporating real-time operational feedback into the prediction model. This feedback loop allows for the continuous refinement of the model, ensuring that it remains responsive to the actual performance of the snow making machines and the prevailing conditions on the ski slope. By dynamically adjusting the predictions, the system can provide more accurate forecasts that better reflect the current and anticipated snow conditions, leading to more effective and efficient snow management operations.

[0048] As used herein, "real-time feedback" denotes the immediate data provided by the snow making machines regarding their operational performance and / or actual data of natural snow, which may in particular include the volume of snow produced, the rate of snow production, and the conditions under which the snow is made. This feedback is used to assess the effectiveness of the snow making process and to make any adjustments to the prediction model that may be necessitated by discrepancies between predicted and actual snow depths.

[0049] In an embodiment of the system for controlling snow making machines, the realtime feedback is provided by sensors deployed on at least one of snow making M / TAL-012-PC

[0050] - 10 - machines, snow groomers, drones, satellites, lift systems, and fixed measurement points on the ski slope.

[0051] The use of sensors at various installation locations enables more accurate and spatially differentiated measurement data, which contributes to improved control of snow production across the slope.

[0052] In an embodiment of the system for controlling snow making machines, the sensors comprise at least one of Light Detection and Ranging (LIDAR) devices, thermal sensors, ultrasonic sensors, ground -penetrating radar, and weather sensors for measuring snow conditions, and optionally GPS units for determining sensor location.

[0053] This configuration facilitates the acquisition of a comprehensive and accurate dataset that encompasses both current snow and weather parameters and the spatial context of the corresponding measurements.

[0054] In an embodiment of the system for controlling snow making machines, the data receiver module is further configured to receive at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slope-specific data related to slopespecific factors. Slope-specific factors include at least one of slope gradient, aspect, and shading patterns. The machine learning module is configured to incorporate the at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slope-specific data into the prediction model. This integration serves to enhance the accuracy of snow depth forecasts.

[0055] This embodiment provides the technical advantage of enabling the prediction model to account for the physical characteristics of the ski slope that directly influence snow accumulation and melting patterns. By incorporating slope-specific factors into the predictive analysis, the system can generate more precise forecasts of snow depth, which are tailored to the individual attributes of each section of the ski slope. M / TAL-012-PC

[0056] - 11 -

[0057] This results in a more efficient allocation of snow management resources and a better skiing experience due to the maintenance of ideal snow conditions.

[0058] The technical advantage of this enhancement is the improved predictive capability of the system, which can now anticipate changes in snow depth due to upcoming weather events. By factoring in future ambient conditions such as temperature fluctuations, precipitation, wind patterns, and other meteorological data, the prediction model can offer more precise and timely forecasts. This proactive approach allows ski slope operators to make data-driven decisions regarding snowmaking and maintenance, ultimately leading to more efficient and effective and even automated snow management.

[0059] As used herein, "weather forecast data" denotes a set of predictions about future weather conditions, typically provided by meteorological services. These forecasts are based on a variety of data sources and models that predict how weather conditions are likely to evolve over a given period. When incorporated into the prediction model for snow depth, these forecasts provide valuable insights that can help to anticipate and respond to the dynamic nature of weather and its impact on snow conditions.

[0060] In an embodiment of the system for controlling snow making machines, the ski slope is divided into a plurality of virtual segments, each virtual segment representing a grid portion of the ski slope for which snow depth is predicted individually. The control module is configured to generate individual control signals for each virtual segment. This division enables localized snow management and enhances the precision of snow depth forecasting.

[0061] The technical advantage of this approach is the ability to manage snow conditions at a granular level, which is particularly beneficial for large ski resorts with varying terrain and microclimates. By focusing on individual segments, the system can provide targeted recommendations for snowmaking and grooming, ensuring that each area of the ski slope is maintained according to its specific requirements. M / TAL-012-PC

[0062] - 12 -

[0063] As used herein, a "virtual segment" denotes a defined area of the ski slope that is treated as a separate entity for the purpose of data analysis and prediction. This segmentation allows for a detailed understanding of the snow depth across the slope, facilitating a more efficient allocation of resources and a better skiing experience. To each segment, a data vector may be assigned, which includes, but is not limited to, snow depth measurements, the geographic position of the segment, and the time of data collection. It may also comprise information on snow density and snow albedo, and may be collected by snow groomers equipped with measurement devices traversing a ski slope. Snow groomers traditionally measure snow depths at the shield (the front blade), but modern groomers use LiDAR to measure snow depths in front of a groomer and also to the sides, thus gaining detailed information of areas around the groomer.

[0064] While for prediction accuracy a very fine grid of segments, and hence a large number of data vectors, may be used, e.g. segments of 20 x 20 cm2or 30 x 30 cm2, the snow production itself is more concerned with what herein is called "zones", a larger area of a slope that may be regarded as large group of virtual segments collectively managed, since neither snow making machines nor snow groomers are able or intended for adjusting the snow height on a segment by segment basis. Combining pluralities of segments to zones, also called zoning, allows for feasible localized snow management, i.e. the zones facilitate the management of snow depth at a scale that is practical for operational purposes, allowing for a more strategic approach to snow management across the ski slope.

[0065] In an embodiment of the system for controlling snow making machines, the control module is configured to generate control signals for a plurality of snow making machines assigned to specific zones of the ski slope corresponding to a group of virtual segments based on the predicted snow depth. The control signals are configured to collectively adjust the snow production rate of the snow making machines and / or adjust snow goals like specified snow amounts for the associated zones of the ski slope. The zones can be managed independently or collectively, allowing for zone-specific or comprehensive snow management. M / TAL-012-PC

[0066] - 13 -

[0067] This embodiment offers the technical advantage of enabling precise and targeted snow production across different areas of the ski slope. By assigning snow making machines to specific zones and adjusting their production rates collectively, the system ensures that each zone receives the appropriate amount of snow based on its individual requirements. This localized control of snow production enhances the efficiency of resource utilization and contributes to a more uniform snow coverage across the ski slope.

[0068] As used herein, "control signals" denote the specific instructions sent to the snow making machines that dictate operational parameters such as the rate of snow production, the direction of snow discharge, and the timing of operations. These signals are derived from the predictive models and are designed to ensure that the snow making machines operate in a manner that aligns with the predicted snow depth and ambient conditions.

[0069] In an embodiment of the system for controlling snow making machines, the control module is configured to generate control signals for a plurality of snow making machines that are mobile and can be moved to different zones on the ski slope based on the predicted snow depth, thereby providing flexibility in snow production.

[0070] This embodiment provides the technical advantage of allowing for dynamic and adaptive snow production in response to changing snow depth requirements across the ski slope. The mobility of the snow making machines enables the system to reposition these resources to areas where they are currently needed the most, ensuring efficient and effective snow coverage throughout the ski slope.

[0071] As used herein, "mobile snow making machines" refer to snow making equipment that is designed to be easily transported from one location to another on the ski slope. This mobility may be facilitated by various means, such as mounted on tracks, wheels, or other transportation mechanisms. A mobile snow making machine may be an unmanned self-driving vehicle, or may be transported to different locations using e.g. snow groomers, helicopters or the like. The control signals in this context include instructions not just for the snow production parameters but also for the movement of the machines to designated zones as dictated by the M / TAL-012-PC

[0072] - 14 - predictive models, e.g. recommendations based on the algorithm as to which mobile machine should be moved where.

[0073] In an embodiment of the system for controlling snow making machines, the control module is configured to generate control signals for the snow making machines based on real-time weather data and real-time energy consumption data to adjust snow production under energy efficiency aspects.

[0074] This configuration allows for dynamic and adaptive snow production in response to changing conditions and promotes energy-efficient operation of the snow making machines.

[0075] The technical advantage of this configuration is the ability to dynamically tailor snow production to the immediate environmental conditions, which can vary rapidly and unpredictably. By utilizing real-time data from weather and energy consumption sensors, the system can make precise adjustments to the snow making machines' operations, such as modifying the water and air mixture or altering the direction and intensity of snow discharge. This responsiveness to real-time conditions leads to more efficient snow production, conserving resources and maintaining the ski slope in a state that is both safe for skiers and conducive to an enjoyable skiing experience.

[0076] As used herein, "real-time weather data" refers to the current environmental factors at the ski slope, such as temperature, humidity, and wind, which can influence the snow making process. "Real-time energy consumption data" denotes the amount of electrical power or other forms of energy used by the snow making machines during the snow production process. The control signals are designed to manage the operations of the snow making machines in a way that balances the snow production requirements with the energy usage, aiming to minimize the overall energy consumption while still achieving the desired snow depth and quality across the group of segments.

[0077] In an embodiment of the system for controlling snow making machines, the control module is configured to generate control signals to direct a subset of the snow M / TAL-012-PC

[0078] - 15 - making machines to produce snow for storage in a snow reservoir. The snow produced for the snow reservoir is intended for later distribution onto the ski slope based on the predicted snow depth and operational requirements of the ski slope.

[0079] This embodiment provides the technical advantage of ensuring a consistent and reliable snow supply by creating a reserve of snow that can be deployed as conditions demand. The ability to store snow and distribute it when it is specifically needed allows for greater flexibility in snow management, particularly during periods of unexpected weather changes or high demand due to increased skier traffic. This proactive approach to snow management can lead to more efficient use of the snow making machines and can help maintain ideal skiing conditions throughout the season.

[0080] As used herein, "snow reservoir" denotes a designated storage area where snow is accumulated and preserved for future use on the ski slope. The snow reservoir allows for strategic management of snow resources, providing a buffer against fluctuations in natural snowfall and enabling the ski resort to respond quickly to varying snow depth requirements.

[0081] The present invention encompasses a method for controlling snow making machines on a ski slope, which includes predicting snow depth on the ski slope using a method comprising receiving historical snow data related to snow depth on the ski slope; training a machine learning algorithm with the historical snow data; ; generating, by the trained machine learning algorithm, a prediction model for forecasting snow depth evolution on the ski slope; receiving current snow data for various locations on the ski slope including measurements of snow depth and associated ambient conditions; applying the current snow data and the current ambient condition data to the prediction model; predicting snow depth on various locations on the ski slope using the prediction model; and generating, based on the predicted snow depth, at least one of a) control signals for automatic transmission to the snow making machines to control at least one of starting snow production, stopping snow production, adjusting snow goals like specified snow amounts, adjusting water flow rates, and adjusting air pressure, and b) control information for presentation on a user interface for manual control of the snow making machines. M / TAL-012-PC

[0082] - 16 -

[0083] The historical snow data includes recorded measurements of snow depth at various locations on the ski slope and associated ambient conditions at the time of each measurement.

[0084] This method has the technical advantage of enabling precise and proactive snow management by utilizing historical and current data to inform snow depth predictions. The use of a machine learning algorithm allows for the analysis of complex patterns in snow depth changes, leading to more accurate forecasts that can guide efficient snowmaking and maintenance decisions. The snowmaking machines may be automatically controlled without requiring manual intervention or alternatively may be manually operated. By outputting actionable recommendations, the method facilitates informed decision-making that can enhance the safety and enjoyment of ski slope users while optimizing resource utilization.

[0085] In an embodiment of the method for controlling snow making machines on ski slopes, the method further comprises receiving real-time feedback regarding actual snow data, i.e. data of snow freshly made by the snow making machines and / or data of natural snow, and continuously updating the prediction model and / or the snow depth prediction based on the real-time feedback.

[0086] The technical advantage of this aspect is the enhanced accuracy of the snow depth predictions, which is achieved by incorporating real-time operational feedback into the prediction model. This feedback loop allows for the continuous refinement of the model, ensuring that it remains responsive to the actual performance of the snow making machines and the prevailing conditions on the ski slope. By dynamically adjusting the predictions, the system can provide more accurate forecasts that better reflect the current and anticipated snow conditions, leading to more effective and efficient snow management operations.

[0087] In an embodiment of the method for controlling snow making machines on ski slopes, the method further comprises receiving at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slope-specific data related to slopespecific factors. Slope-specific factors include at least one of slope gradient, aspect, M / TAL-012-PC

[0088] - 17 - and shading patterns. Furthermore, the method comprises incorporating the at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slopespecific data into the prediction model. This integration serves to enhance the accuracy of snow depth forecasts.

[0089] The technical advantage of this enhancement is the improved predictive capability of the system, which can now anticipate changes in snow depth due to upcoming weather events. By factoring in future ambient conditions such as temperature fluctuations, precipitation, wind patterns, and other meteorological data, the prediction model can offer more precise and timely forecasts. This proactive approach allows ski slope operators to make data-driven decisions regarding snowmaking and maintenance, ultimately leading to more efficient and effective and even automated snow management.

[0090] This embodiment provides the technical advantage of enabling the prediction model to account for the physical characteristics of the ski slope that directly influence snow accumulation and melting patterns. By incorporating slope-specific factors into the predictive analysis, the system can generate more precise forecasts of snow depth, which are tailored to the individual attributes of each section of the ski slope. This results in a more efficient allocation of snow management resources and a better skiing experience due to the maintenance of ideal snow conditions.

[0091] In an embodiment of the method for controlling snow making machines on ski slopes, the ski slope is divided into a plurality of virtual segments representing a grid portion of the ski slope for which snow depth is predicted individually. The method comprises generating individual control signals for each virtual segment. This division enables localized snow management and enhances the precision of snow depth forecasting.

[0092] The technical advantage of this approach is the ability to manage snow conditions at a granular level, which is particularly beneficial for large ski resorts with varying terrain and microclimates. By focusing on individual segments, the system can M / TAL-012-PC

[0093] - 18 - provide targeted recommendations for snowmaking and grooming, ensuring that each area of the ski slope is maintained according to its specific requirements.

[0094] In an embodiment of the method for controlling snow making machines on ski slopes, the method further comprises generating control signals for a plurality of snow making machines assigned to specific zones of the ski slope corresponding to a group of virtual segments based on the predicted snow depth. The control signals collectively adjust the snow production rate of the snow making machines and / or adjust snow goals like specified snow amounts for the associated zones of the ski slope. The zones are managed independently or collectively for zone-specific or comprehensive snow management.

[0095] This embodiment offers the technical advantage of enabling precise and targeted snow production across different areas of the ski slope. By assigning snow making machines to specific zones and adjusting their production rates collectively, the system ensures that each zone receives the appropriate amount of snow based on its individual requirements. This localized control of snow production enhances the efficiency of resource utilization and contributes to a more uniform snow coverage across the ski slope.

[0096] In an embodiment of the method for controlling snow making machines on ski slopes, the method further comprises generating control signals for a plurality of snow making machines that are mobile and repositioning the mobile snow making machines to different zones on the ski slope based on the predicted snow depth, thereby providing flexibility in snow production.

[0097] This embodiment provides the technical advantage of allowing for dynamic and adaptive snow production in response to changing snow depth requirements across the ski slope. The mobility of the snow making machines enables the system to reposition these resources to areas where they are currently needed the most, ensuring efficient and effective snow coverage throughout the ski slope.

[0098] In an embodiment of the method for controlling snow making machines on ski slopes, the method further comprises generating control signals for the snow M / TAL-012-PC

[0099] - 19 - making machines based on real-time weather data and real-time energy consumption data to adjust snow production under energy efficiency aspects.

[0100] This configuration allows for dynamic and adaptive snow production in response to changing conditions and promotes energy-efficient operation of the snow making machines.

[0101] The technical advantage of this configuration is the ability to dynamically tailor snow production to the immediate environmental conditions, which can vary rapidly and unpredictably. By utilizing real-time data from weather and energy consumption sensors, the system can make precise adjustments to the snow making machines' operations, such as modifying the water and air mixture or altering the direction and intensity of snow discharge. This responsiveness to real-time conditions leads to more efficient snow production, conserving resources and maintaining the ski slope in a state that is both safe for skiers and conducive to an enjoyable skiing experience.

[0102] In an embodiment of the method for controlling snow making machines on ski slopes, the method comprises generating control signals to direct a subset of the snow making machines to produce snow for storage in a snow reservoir. The snow produced for the snow reservoir is intended for later distribution onto the ski slope based on the predicted snow depth and operational requirements of the ski slope.

[0103] This embodiment provides the technical advantage of ensuring a consistent and reliable snow supply by creating a reserve of snow that can be deployed as conditions demand. The ability to store snow and distribute it when it is specifically needed allows for greater flexibility in snow management, particularly during periods of unexpected weather changes or high demand due to increased skier traffic. This proactive approach to snow management can lead to more efficient use of the snow making machines and can help maintain ideal skiing conditions throughout the season.

[0104] In an embodiment of the method for controlling snow making machines on ski slopes, the ambient conditions may include at least temperature, and optionally M / TAL-012-PC

[0105] - 20 - additional weather data such as hours of sunshine, wind speed and direction, and humidity. This comprehensive consideration of ambient conditions enhances the accuracy of snow depth predictions.

[0106] The technical advantage of incorporating a broad range of ambient conditions lies in the system's ability to account for the multifaceted nature of weather and its impact on snow depth. Temperature, sunshine, wind, and humidity all play a role in the rate of snow accumulation and melt. By including these variables, the prediction model can more accurately simulate the real-world conditions affecting the ski slope, leading to more reliable snow management decisions.

[0107] As used herein, "ambient conditions" refer to the environmental factors present at the ski slope that can influence snow depth. These conditions provide contextual data that informs the prediction model, ensuring that forecasts are reflective of the actual climate and weather patterns experienced at the ski slope.

[0108] In an embodiment of the method for controlling snow making machines on ski slopes, the method may comprise outputting information on at least one of expected snow conditions, potential areas of snow deficiency, and advisories for snowmaking. Furthermore, the method may comprise outputting recommendations related to at least one of specific actions for snow management. These specific actions include initiating snowmaking operations, adjusting snowmaking parameters, and redistributing snow on the ski slope.

[0109] This embodiment provides the technical advantage of delivering actionable insights for ski slope operators, enabling them to respond effectively to changing snow conditions. By providing detailed information on expected snow conditions and potential deficiencies, the system allows for proactive snow management. The inclusion of advisories and recommendations ensures that operators have the guidance they require to maintain the ski slope in an optimum state, whether that involves initiating additional snowmaking, adjusting existing snowmaking operations, or redistributing snow to maintain even coverage. M / TAL-012-PC

[0110] - 21 -

[0111] "Snowmaking parameters" refer to the controllable variables of snowmaking machines, such as water flow rate, air pressure, and snowing direction, which can be adjusted to influence the quantity and quality of artificial snow produced.

[0112] In an embodiment of the method for controlling snow making machines on ski slopes, the method may comprise receiving data related to slope-specific factors that do not change over time, and wherein the machine learning algorithm is further trained based on the slope-specific factors. This enables the machine learning algorithm to extrapolate and output information and predictions for slopes with limited or no historical snow data by leveraging learned patterns from slopes with comprehensive historical data. Furthermore, the machine learning algorithm may be configured to apply the prediction model to a plurality of ski slopes, each with potentially different slope-specific factors, to predict snow depth across multiple slopes. This enables a comprehensive snow management strategy for a ski resort with diverse terrain.

[0113] This approach offers the technical advantage of enhancing the predictive capabilities of the system for ski slopes that may not have extensive historical snow depth records. By incorporating slope-specific factors into the training of the machine learning algorithm, the system can apply learned patterns from well-documented slopes and zones / virtual segments to those with sparse data, ensuring that accurate predictions can still be made. This enables ski resort operators to manage snow conditions effectively, even on slopes that are new or have not been monitored extensively in the past.

[0114] As used herein, "slope-specific factors" refer to the inherent characteristics of the ski slope that remain constant over time and affect snow behavior. These factors include the gradient or steepness of the slope, the aspect or compass orientation which determines sun exposure, and shading patterns that may result from surrounding topography or structures. When integrated in the training of the machine learning algorithm, these factors provide context for understanding how snow depth evolves in response to environmental conditions. M / TAL-012-PC

[0115] - 22 -

[0116] By considering these constant factors, the algorithm can make informed predictions about snow depth evolution, which is particularly useful for slopes that lack a detailed historical data record, allowing the trained model to be used on a plurality of slopes even of a resort, for which only slope-specific factors are known. Accordingly, the machine learning algorithm may be configured to apply the prediction model to a plurality of ski slopes, each with potentially different slopespecific factors, to predict snow depth across multiple slopes, thereby enabling a comprehensive snow management strategy for a ski resort with diverse terrain.

[0117] In an embodiment of the method for controlling snow making machines on ski slopes, the method comprises generating optimistic and pessimistic predictions of snow depth for zones of the ski slope, wherein the optimistic predictions assume favorable weather conditions for snow retention and the pessimistic predictions assume adverse weather conditions. This provides a range of potential snow depth outcomes.

[0118] "Optimistic predictions" and "pessimistic predictions" denote the range of forecasts generated by the system, reflecting the potential outcomes based on varying weather conditions, which are used to inform snow management strategies and ensure the ski slope is prepared for different weather-related impacts on snow depth. Optimistic predictions, which assume favorable weather conditions for snow retention, can guide decisions for scenarios where less intervention may be sufficient. Conversely, pessimistic predictions, which assume adverse weather conditions, help operators prepare for worst-case scenarios, ensuring that proactive measures can be taken to maintain adequate snow coverage.

[0119] The technical advantage of this feature is the ability to prepare for a variety of potential future scenarios, allowing ski slope operators to make informed decisions that account for the full spectrum of possible weather events. Note that the invention, as will be explained later, also allows fully automated snow management, so that "informed decisions" will be taken automatically, and semi -automated snow management guided by a choice of a certain snowmaking strategy by the operator (e.g. "whatever it takes" strategy vs. "energy-saving-strategy"). M / TAL-012-PC

[0120] - 23 -

[0121] In an embodiment of the method for controlling snow making machines on ski slopes, the method comprises generating short-term and long-term predictions of snow depth for zones of the ski slope based on the trained machine learning algorithm.

[0122] These predictions are formulated to provide a range of potential snow depth outcomes, enabling ski resort operators to engage in robust snow management planning.

[0123] As used herein, the term "short-term" refers to a time frame that typically encompasses the immediate future, generally ranging from the present moment up to a few days ahead. In the context of snow management, short-term predictions involve forecasting snow depth and conditions that are expected to occur in the near future, such as within the next 24 to 72 hours. These predictions are particularly useful for making operational decisions that require a rapid response to changing weather conditions and snow levels on the ski slopes.

[0124] Conversely, the term "long-term" pertains to a more extended time frame, which can span from several days, typically about 10 days, to weeks or even months into the future. Long-term predictions in snow management are used to forecast trends in snow depth and conditions over a prolonged period. These forecasts are valuable for strategic planning purposes, such as scheduling snowmaking operations, managing resources over the course of a ski season, and preparing for events that are planned well in advance.

[0125] The short-term predictions offer ski resort operators the ability to make immediate decisions regarding snowmaking, grooming, and slope maintenance, ensuring that the ski slopes are in the desired condition for the immediate future. This is particularly beneficial for addressing sudden changes in weather that can affect snow conditions, such as unexpected snowfall or a sudden rise in temperature.

[0126] On the other hand, the long-term predictions allow operators to develop comprehensive strategies for the upcoming weeks and months. These strategies may include determining the ideal times for snow production to build up the base M / TAL-012-PC

[0127] - 24 - layer of snow, planning for the conservation of snow during anticipated warmer periods, and scheduling major events or competitions with confidence in the availability of suitable snow conditions. In addition, scenarios for the expansion of the snowmaking system can be simulated, e.g. conversion to a more powerful snow making machine.

[0128] The technical advantage of providing both short-term and long-term predictions lies in the ability to support a wide range of operational and strategic decisions. By offering a spectrum of forecasts, the system enables ski resort operators to optimize their snow management practices, ensuring both the immediate and future quality of the ski slopes while also promoting efficient use of resources and maintaining safety standards for skiers.

[0129] One embodiment of the method for controlling snow making machines on a ski slope includes the steps of receiving weather forecast data that provide predictions of future ambient conditions affecting the ski slope, and incorporating the weather forecast data into the prediction model. This integration serves to enhance the accuracy of snow depth forecasts.

[0130] The technical advantage of this enhancement is the improved predictive capability of the system, which can now anticipate changes in snow depth due to upcoming weather events. By factoring in future ambient conditions such as temperature fluctuations, precipitation, wind patterns, and other meteorological data, the prediction model can offer more precise and timely forecasts. This proactive approach allows ski slope operators to make data-driven decisions regarding snowmaking and maintenance, ultimately leading to more efficient and effective and even automated snow management.

[0131] In an embodiment of the method for controlling snow making machines, the method further comprises configuring the control signals to adjust the operations of the snow making machines based on real-time ambient conditions and feedback from sensors located at the ski slope, particularly sensors situated in the snow making machines. This configuration enables the optimization of snow production in M / TAL-012-PC

[0132] - 25 - response to changing weather patterns and ensures that the snow depth within the group of virtual segments remains within a predetermined range.

[0133] The technical advantage of this configuration is the ability to dynamically tailor snow production to the immediate environmental conditions, which can vary rapidly. By utilizing real-time data from sensors, the system can make precise adjustments to the snow making machines' operations, such as modifying the water and air mixture or altering the direction and intensity of snow discharge. This responsiveness to real-time conditions leads to more efficient snow production, conserving resources and maintaining the ski slope in a state that is both safe for skiers and conducive to an enjoyable skiing experience.

[0134] As used herein, "real-time ambient conditions" refer to the current environmental factors at the ski slope, such as temperature, humidity, and wind, which can influence the snow making process. "Feedback from sensors" denotes the data collected by sensors integrated into the snow making machines, which provide information on operational parameters and the effectiveness of the snow production. This feedback is instrumental in the adaptive control of the snow making machines, ensuring that snow depth targets are met and maintained within the desired range for each virtual segment of the ski slope.

[0135] In an embodiment of the method for controlling snow making machines, the method further comprises adapting the control signals to initiate or cease snow production by the snow making machines based on a threshold snow depth value determined for the group of virtual segments. Additionally, the control signals are in this embodiment configured to modulate the mixture of water and air in the snow making machines to produce snow of a desired quality and consistency for the group of segments.

[0136] This embodiment provides the technical advantage of precise control over the snowmaking process, ensuring that snow is produced when and where it is actually needed, and to the quality standards that are desired for the ski slope. By setting threshold values for snow depth, the system can automatically start or stop snow production, thereby preventing overproduction and conserving resources. The ability M / TAL-012-PC

[0137] - 26 - to modulate the water and air mixture allows for the creation of snow that meets specific requirements, such as wetter snow for base layers or drier snow for top layers, enhancing the skiing experience and safety on the slopes.

[0138] As used herein, "threshold snow depth value" refers to a predefined measurement of snow depth that triggers the snow making machines to either start or stop producing snow. This value is determined based on the requirements of the ski slope and the individual virtual segments within it. "Modulate the mixture of water and air" denotes the adjustment of the proportions of water and compressed air used in the snow making process, which affects the characteristics of the artificial snow produced, such as its moisture content, density, and texture. These adjustments are made through the control signals to ensure that the snow produced matches the desired quality and consistency for each segment of the ski slope.

[0139] In an embodiment of the method for controlling snow making machines, the method further comprises generating control signals in consideration of real-time energy consumption, thereby promoting energy-efficient operation of the snow making machines across the group of segments.

[0140] This embodiment has the technical advantage of promoting sustainable and cost- effective snow management practices by optimizing the energy usage of the snow making machines. By factoring in energy consumption when generating control signals, the system ensures that snow production is not just effective but also aligns with energy conservation goals. This can lead to a reduction in the environmental impact of snowmaking operations and can result in lower operational costs for ski resorts.

[0141] As used herein, "energy consumption" refers to the amount of electrical power or other forms of energy used by the snow making machines during the snow production process. The control signals are designed to manage the operations of the snow making machines in a way that balances the snow production requirements with the energy usage, aiming to minimize the overall energy consumption while still achieving the desired snow depth and quality across the group of segments. M / TAL-012-PC

[0142] - 27 -

[0143] In an embodiment of the system for controlling snow making machines, the data receiver module is further configured to receive data related to slope-specific factors, which include at least one of slope gradient, aspect, and shading patterns. These factors are utilized by the machine learning module to enhance the accuracy of the prediction model.

[0144] This embodiment provides the technical advantage of enabling the prediction model to account for the physical characteristics of the ski slope that directly influence snow accumulation and melting patterns. By incorporating slope-specific factors into the predictive analysis, the system can generate more precise forecasts of snow depth, which are tailored to the individual attributes of each section of the ski slope. This results in a more efficient allocation of snow management resources and a better skiing experience due to the maintenance of ideal snow conditions.

[0145] In an embodiment of the system for controlling snow making machines, the output module is configured to provide a graphical representation of the predicted snow depth over time for various locations on the ski slope. This feature enables a visual comparison of current predictions with historical snow data.

[0146] The technical advantage of this embodiment is the enhanced ability for ski slope operators to visually assess and understand the snow depth trends and make informed decisions based on a clear graphical representation. By comparing current predictions with historical snow data, operators can quickly identify patterns, anomalies, or deviations from expected snow depth levels, facilitating proactive snow management actions. This visual approach simplifies the interpretation of complex data, making it more accessible and actionable for those responsible for maintaining the ski slope.

[0147] As used herein, "graphical representation" refers to a visual depiction of data, such as charts, graphs, or maps, that illustrates the predicted snow depth across different locations on the ski slope over a specified time period. This representation can include various visual elements such as lines, bars, color gradients, or other M / TAL-012-PC

[0148] - 28 - graphical indicators that convey the magnitude and distribution of snow depth, enabling a straightforward comparison with historical snow depth measurements.

[0149] In an embodiment of the system for controlling snow making machines, the machine learning module is further configured to update the prediction model dynamically based on feedback received from the snow making machines, snow grooming machines or other sources regarding the actual snow depth after snow making operations. This dynamic updating process serves to improve the predictive accuracy of the system over time.

[0150] The technical advantage of this feature is the ability to refine the prediction model continuously, ensuring that it remains responsive to the actual conditions on the ski slope. By incorporating feedback from the snow making machines, snow grooming machines or other sources, which includes data on the efficacy of snow production, the machine learning module can adjust the model to more accurately reflect the observed snow depths. This feedback loop enhances the reliability of the system's predictions, leading to more effective and efficient snow management.

[0151] In an embodiment of the system for controlling snow making machines, the data receiver module is further configured to receive weather forecast data predicting future ambient conditions, and the machine learning module is configured to incorporate the weather forecast data into the prediction model to improve the predictive accuracy of future snow depth on the ski slope.

[0152] This embodiment provides the technical advantage of enhancing the system's predictive capabilities by integrating anticipated environmental changes into the snow depth forecasts. The inclusion of weather forecast data allows the machine learning module to adjust the prediction model proactively, taking into account expected temperature fluctuations, precipitation, wind patterns, and other meteorological data that can influence snow accumulation and melting. This proactive approach to incorporating future ambient conditions into the prediction model leads to more accurate and timely snow depth forecasts, enabling ski slope operators to make better-informed decisions regarding snowmaking and maintenance operations. M / TAL-012-PC

[0153] - 29 -

[0154] The present invention encompasses a system for controlling snow making machines on a ski slope comprising the a system for predicting snow depth on a ski slope and including a control module that is communicatively coupled to snow making machines and that is configured to generate control signals based on the predicted snow depth. The snow making machines are configured to receive the control signals from the control module and adjust snow production in response to the control signals.

[0155] The technical advantage of this embodiment is the automation of snow making operations, which leads to a more efficient and responsive system for controlling snow making machines. By utilizing predictive models to inform the control of snow making machines, the system can dynamically adjust snow production to match the actual snow requirements of the ski slope. This results in optimized resource usage, reduced waste, and ensures that the ski slope maintains the desired snow conditions for safety and enjoyment.

[0156] A system for controlling snow making machines offers the technical advantage of integrating advanced control capabilities with a fleet of snow making machines to optimize snow production on ski slopes. The control system is configured to generate control signals based on predictive models of snow depth, real-time weather data, and energy consumption data, thereby enabling dynamic and adaptive snow production. This integration promotes energy-efficient operation of the snow making machines, ensuring that snow is produced and distributed according to the specific requirements of the ski slope while minimizing resource waste.

[0157] In a further advantageous embodiment, the system and the method may be configured to receive real-time energy pricing data from power plants or energy suppliers, enabling cost-optimized snow production scheduling. The data receiver module may be configured to receive current and forecasted energy prices, peak demand periods, and availability of renewable energy sources. The control module can then incorporate this pricing information when generating control signals, allowing the system to schedule intensive snow production during periods of lower M / TAL-012-PC

[0158] - 30 - energy costs while reducing or postponing snow production during peak pricing periods. This energy cost optimization may be balanced against the predicted snow depth requirements and weather forecasts to ensure adequate snow coverage while minimizing operational costs. For example, if energy prices are expected to be significantly lower during overnight hours and weather conditions are favorable for snow production, the system may generate control signals to increase snow production during these cost-effective periods, even if immediate snow depth requirements are already met, thereby building up snow reserves in the snow reservoir for later distribution. Conversely, during periods of high energy costs, the system may prioritize only essential snow production to maintain minimum required snow depths. This approach enables ski resorts to significantly reduce their energy expenses while maintaining optimal snow conditions through intelligent scheduling of snow production activities.

[0159] Further aspects, details and advantages of the invention will become apparent from the following detailed description of preferred embodiments in conjunction with the drawing, which comprises two figures.

[0160] BRIEF DESCRIPTION OF THE DRAWING

[0161] Fig. 1 is a schematic representation of a system for controlling snow making machines, according to aspects of the present disclosure.

[0162] Fig. 2 illustrates a ski slope equipped with the system for controlling snow making machines of Fig. 1, according to aspects of the present disclosure.

[0163] DESCRIPTION OF PREFERRED EMBODIMENTS

[0164] Fig. 1 depicts very schematically a system for controlling snow making machines according to the invention, denoted in its entirety by 10, engineered to enhance snow production and maintenance on ski slopes by harnessing advanced data M / TAL-012-PC

[0165] - 31 - processing and machine learning techniques. The system 10 consists in this embodiment of a network of modules, each with a distinct role in facilitating efficient snow management.

[0166] The system for controlling snow making machines 10 is an advanced framework designed to optimize snow production and maintenance on ski slopes. At the core of the system is a data processing unit 12, which serves as the central hub for analyzing data inputs, processing this information, and generating actionable outputs for effective snow management across one or multiple ski slopes.

[0167] A data storage unit 14 archives historical snow data, which includes detailed measurements of snow depth at various locations on the ski slopes and the ambient conditions at the time of each measurement.

[0168] Environmental sensors 16 are deployed across the ski slopes to gather real-time data on ambient conditions such as temperature, humidity, wind speed, and direction. This real-time data, along with additional information from external data sources 18, the internet 20 and the data storage unit 14 is funneled to the data processing unit 12, providing a comprehensive dataset that reflects both current conditions and predictive insights for future snow management. Such dataset may in particular comprise

[0169] 1. Historical Data: This is a primary input to the machine learning algorithm. It includes data from several seasons for each slope. The historical data is used to train the machine learning algorithm to predict the evolution of snow depth.

[0170] 2. Slope-Specific Factors: These may include parameters that are specific to each slope and do not change over time. They may include slope angle (gradient or steepness), aspect or compass orientation, which determines sun exposure, and shading patterns that may result from surrounding topography or structures. They may also include slope-specific variables that change over time like number of skiers and can be used for both training the machine learning algorithm and generating the predictions. M / TAL-012-PC

[0171] - 32 -

[0172] 3. Forecasts: Various forecasts may be incorporated into prediction to increase its accuracy. These forecasts may include weather forecast and number of skiers forecast and can in particular be for generating the predictions.

[0173] 4. Current Snow Data: This data is typically collected by snow groomers that measure snow depth. It may include various further metrics related to snow such as snow density and snow albedo (the ability of snow to reflect rather than to absorb sunlight; if the snow is 'dirty' from some reason, more light is absorbed and the snow melts faster) as well as information on external conditions affecting snow depth such as temperature, wind, sun, and number of skiers. These parameters can be used for both training the machine learning algorithm and generating the predictions. For example, if the snow on the respective ski slope has a high albedo, the system may predict a slower rate of snow melt and a higher future snow depth.

[0174] 5. Data from Other Resorts: In some embodiments, data from other, in particular near-by resorts is taken into account. For example, a shutdown of slopes or lifts in a neighboring resort may increase the number of skiers in the present resort. This data can be used for both training the machine learning algorithm and generating the predictions.

[0175] A data receiving module 22 is responsible for the initial collection and organization of incoming data from the environmental sensors 16 and external data sources 18. This module ensures that all data is accurately captured and made available for further processing.

[0176] A machine learning module 24 is a core component of the system, which employs sophisticated algorithms to discern patterns and train a predictive model for forecasting snow depth evolution. As outlined above, the snow data, current and historical, may be arranged as a "grid" of data vectors assigned to specific positions on a ski slope. The machine learning module 24 utilizing the snow data to train a machine learning algorithm. This algorithm is adept at generating a prediction model that can accurately forecast snow depth evolution on the ski slopes.

[0177] Importantly, the machine learning module 24 is also capable of extrapolating information for slopes with limited or no historical snow depth data by leveraging M / TAL-012-PC

[0178] - 33 - learned patterns from slopes with comprehensive historical data. This feature is particularly beneficial for ski resorts that are newly established or have not previously collected extensive snow data, or new slopes or slope sections within existing ski resorts.

[0179] An application module 26 applies the current snow data to the prediction model, ensuring that the forecasts are up-to-date and reflective of the latest conditions. It integrates real-time or recently collected snow depth data and current ambient condition data into the prediction model. This step is integral to ensuring that the forecasts for snow depth are accurate and reflective of the latest conditions on the respective ski slope, for which a prediction shall be made.

[0180] Upon receiving current snow data, which includes measurements of snow depth and associated ambient conditions for various locations on the respective ski slope, the application module 26 in this embodiment performs the following functions:

[0181] 1. Data Integration: The application module 26 integrates the current snow data with the historical snow data stored in the data storage unit 14. This integration allows the machine learning module 24 to consider both past and present conditions when generating the prediction model.

[0182] 2. Model Updating: The application module 26 updates the prediction model with the current snow data. This may involve recalibrating the model parameters to reflect the new data inputs, ensuring that the model remains sensitive to recent changes in snow depth and ambient conditions.

[0183] 3. Real-Time Analysis: The application module 26 analyzes the current snow data in real-time, comparing it against the predictions made by the prediction model. Discrepancies between the predicted and actual snow depths are used to fine-tune the model, improving its predictive accuracy.

[0184] 4. Feedback Incorporation: The application module 26 may incorporate feedback in particular from snow making machines and environmental sensors 16. This M / TAL-012-PC

[0185] - 34 - feedback includes information on the efficacy of snow production and the current state of the snowpack, which is used to adjust the prediction model accordingly.

[0186] A prediction module 28 is responsible for predicting the snow depth on the ski slopes using the prediction model. It processes the data to generate forecasts for future snow depth, which are based on the learned patterns from the historical data and the current conditions applied to the model. In this exemplary configuration, the application module 26 and the prediction module 28 are distinct components within the system for controlling snow making machines 10, each serving a specific function in the process of managing snow depth on a ski slope:

[0187] The application module 26 is responsible for applying the current snow depth data and the current ambient condition data to the prediction model. Essentially, it integrates real-time data into the model to ensure that the predictions are based on the latest available information. This module acts as an intermediary that processes and updates the prediction model with new data inputs, allowing the system to maintain an accurate and current forecast of snow depth conditions.

[0188] The prediction module 28 is tasked with predicting snow depth on the ski slope using the prediction model. This module takes the refined prediction model, which has been updated by the application module 26 with the latest data, and uses it to generate forecasts for future snow depth across various locations on the ski slope. The prediction module 28 is the component that actually performs the analysis and generates the snow depth predictions that will be used to inform snow management decisions.

[0189] In other words: the application module 26 prepares and updates the prediction model with current data, while the prediction module 28 uses the updated model to make predictions about future snow depth. Both modules work in tandem to ensure that the system for controlling snow making machines 10 can provide accurate and actionable forecasts for effective snow management on the ski slopes.

[0190] By applying the current snow data in this manner, the system for controlling snow making machines 10 ensures that the prediction model is continuously refined and adapted to the dynamic conditions of the respective ski slope. This results in more reliable and actionable forecasts, which in turn inform the recommendations and M / TAL-012-PC

[0191] - 35 - control signals generated by an output module 30 and a control module 32, respectively.

[0192] The system for controlling snow making machines 10 in this embodiment is designed to generate optimistic and pessimistic predictions as well as short-term and long-term predictions of snow depth for zones of a respective ski slope based on the trained machine learning algorithm. For this purpose, the machine learning module 24 employs advanced algorithms to analyze historical snow data, current conditions, and a variety of predictive inputs, such as weather forecasts and skier traffic estimates. Based on this analysis, the prediction module 28 utilizes the prediction model to generate four sets of predictions:

[0193] 1. Optimistic Predictions: These are based on the assumption of favorable weather conditions for snow retention, such as lower temperatures and the absence of rain or intense sunlight. The system also considers the possibility of lower-than-average skier traffic, which would result in less snow consumption and degradation.

[0194] Optimistic predictions guide decisions for scenarios where less intervention in terms of snowmaking or snow conservation efforts may be sufficient.

[0195] 2. Pessimistic Predictions: Conversely, these predictions assume adverse weather conditions that could lead to accelerated snow melt or suboptimal snow retention, such as higher temperatures or precipitation. Additionally, the system accounts for the potential impact of higher-than-average skier traffic, which can reduce snow depth more quickly due to increased consumption and displacement. Pessimistic predictions help ski resort operators prepare for worst-case scenarios, ensuring that proactive measures can be taken to maintain adequate snow coverage.

[0196] 3. Short-Term Predictions: These predictions focus on the upcoming days and are particularly useful for immediate operational decisions. They enable ski resort operators to respond quickly to imminent changes in weather and snow conditions, ensuring that the slopes are well-prepared for daily skier traffic and any near-term events. M / TAL-012-PC

[0197] - 36 -

[0198] 4. Long-Term Predictions: These predictions extend further into the future, providing insights for strategic planning over the course of the ski season. Longterm predictions are valuable for scheduling snow production, anticipating periods of high skier traffic, and preparing for major events that require consistent and reliable snow conditions.

[0199] By preparing for both optimistic and pessimistic scenarios, the system for controlling snow making machines 10 ensures that ski resorts can maintain high-quality slopes and provide a consistent skiing experience, regardless of the variability in weather conditions and skier traffic. This dual-prediction approach also contributes to the efficient use of resources, as it allows for targeted snowmaking that aligns with the specific forecasted conditions for each zone of the ski slope.

[0200] The output module 30 may then communicate these predictions to the ski resort and serves as the communication interface between the system and a ski slope operator. It outputs information and recommendations based on the predicted snow depth, providing actionable insights for snow management and a comprehensive understanding of the potential range of snow depth outcomes. This information enables operators to make informed decisions about snowmaking and slope maintenance, such as whether to increase snow production in anticipation of potential snow deficits or to conserve resources when conditions are expected to be favorable. This information also enables a fully automated control of snow making machines.

[0201] In the shown embodiment, the control module 32 is configured to interpret the predictions from the prediction module 28 and generate control signals for snow making machines. These signals dictate the operational parameters such as the rate of snow production, the direction of snow discharge, and the timing of operations.

[0202] In some embodiments, the control module 32 may also be configured to consider real-time operational data, which could include energy consumption data, to further optimize the snow production process. This consideration of energy consumption data would allow the system to promote energy-efficient operation of the snow M / TAL-012-PC

[0203] - 37 - making machines, adjusting their activity not just for the predicted snow depth and ambient conditions, but also for the energy usage patterns observed in real-time.

[0204] An output interface 34 allows for the display of system outputs and a user interface 36 enables user interaction. Operators can view predictions, control signals, and recommendations, and can input commands or adjustments as deemed appropriate. Note that output interface 34 and user interface 36 may be integrated in a single device, like a touch screen or a mobile communication device like a mobile phone or tablet computer.

[0205] The control unit 38 is the link between the data processing unit 12 and snow making machines 40, some of which are shown in Fig. 1. In this example, each snow making machines 40 is equipment with a sensor array 42 comprising different sensors for providing real-time information on operational parameters of the respective snow making machine 40. For sake of clarity, only some machines 40 and some sensor arrays 42 have been provided with reference numbers. In practice, snow making machines may be employed not having weather sensors on board may be used. In such case, weather data from other sources like nearby snow making machines with respective sensors or stand-alone weather stations may be assigned to these machines remotely. Control unit 38 sends the control signals to the snow making machines 40 communicatively coupled to the control unit 38. The sensors in the sensor arrays 42 provide real-time feedback on the snow making process, thus allowing to close the loop and enabling the system to adapt and refine its predictions and control strategies continuously.

[0206] If the system is to incorporate real-time energy consumption data, the following steps may be included:

[0207] 1. Energy Data Acquisition: The snow making machines 40, equipped with sensors, would measure the energy consumed during snow production and transmit this data to the control unit 38. M / TAL-012-PC

[0208] - 38 -

[0209] 2. Energy Efficiency Analysis: The control module 32 would analyze the energy consumption data in conjunction with the snow depth predictions and ambient condition data to identify opportunities for energy savings.

[0210] 3. Control Signal Adjustment: Based on this analysis, the control module 32 would adjust the control signals sent to the snow making machines 40 to optimize their operation for energy efficiency without compromising the desired snow depth and quality.

[0211] 4. Feedback Loop: The system would use feedback from the snow making machines 40 to continuously refine the energy efficiency of the snow production process, learning from past performance to improve future operations.

[0212] By incorporating real-time energy consumption data into the control strategy, the system for controlling snow making machines 10 can ensure that snow production is not just effective in maintaining the desired snow depth, but also conducted in an environmentally responsible and cost-effective manner.

[0213] FIG. 2 illustrates the practical application of the system for controlling snow making machines 10 on a ski slope 44, showcasing the system's advanced capabilities in managing and maintaining ideal snow conditions. The ski slope 44 is virtually divided into a plurality segments 46 and zones 48 (only six zones being depicted in Fig. 2), as explained above. For sake of clarity, only some zones and segments have been provided with reference numbers. The virtual segments 46 may correspond to data collection points, i.e. points, about which data is present for example data collected by a snow groomer as it traverses the ski slope 44. As already mentioned above, snow groomers can be fitted with GPS units that track their precise location as they move, and have typically ground-penetrating radar, LIDAR, or ultrasonic sensors that measure the depth of the snow beneath them. These sensors can provide real-time data on snow depth, density, and surface conditions.

[0214] As the groomer moves across the ski slope 44, it continuously collects data points at regular intervals. Each data point typically includes the groomer's location (latitude, longitude, and altitude), the snow depth at that location, and other relevant ambient conditions such as temperature. The ski slope 44 is virtually divided into M / TAL-012-PC

[0215] - 39 - segments 46, which are smaller areas of the ski slope 44 and can be defined in various shapes and sizes, such as squares or rectangles, depending on the resolution of data desired. Each segment represents a specific area where snow conditions are to be measured. As the groomer passes over or by a segment 46, it collects multiple data points that correspond to that segment 46.

[0216] The collection of data points within a segment 46 provides a detailed picture of the snow conditions in that specific area 48. For example, if a segment 46 is defined as a 15 x 15 centimeter square, the groomer might collect data points at each corner and several points in between, depending on the precision of the sensors and the granularity of data collection desired. The collected data points are then analyzed to determine the average snow depth within each segment 46, variations in snow density, and other characteristics. This information is used to create a detailed map of snow conditions across the ski slope 44. The data collected by the groomer is fed into the system for controlling snow making machines 10, where it is used to update the prediction model for snow depth. This updated model can then inform decisions for snowmaking, grooming, and other slope maintenance activities to ensure the desired snow conditions are achieved for each segment 46.

[0217] By collecting data points that correspond to predefined segments 46, groomers enable the system for controlling snow making machines 10 to accurately predict and control snow conditions at a granular level, leading to more efficient and effective slope management.

[0218] The snow making machines 40 are strategically deployed along the ski slope 44, with each machine in this example responsible for a specific snow making area 50 (only two of which are depicted in Fig. 2). The snow making areas 50 may overlap, and in use, depending on wind conditions, the majority of snow produced by a respective snow making machine 40 should land in a snow making area 50 to which the machine is assigned. Depending on the type of snow making machine and the production rate, it may be foreseen that snow produced in a snow making area 50 is further distributed by groomers, which may also be communicatively coupled to the system for controlling snow making machines 10, for example to receive specific instructions in a fully automated manner. Some types of snow making machines 40, M / TAL-012-PC

[0219] - 40 - in particular so-called lances, may very evenly spray snow on a ski slope such that a snow groomer is not required for the distribution of the snow made by these machines. This works particularly well for so-called "pull-paths", i.e. rather flat parts of ski slopes.

[0220] The snow making machines 40 may be implemented in various configurations to suit the specific requirements of the ski slope 44. These configurations may include stationary snow making machines 40 that are fixed in location, like in particular lance-type snow making machines, or mobile snow making machines 40 that can be moved to different areas of the ski slope 44 as dictated by the predictive models. Each type of snow making machine 40 is strategically utilized within the system to optimize snow production and distribution, ensuring that the ski slope 44 maintains consistent and high-quality snow coverage. Typically, each snow making machine 40 is equipped with a sensor array 42, the sensors of which continuously monitor the snow production process, gathering data on variables such as snow volume, water temperature, and ambient environmental conditions. This data facilitates the aforementioned feedback loop, enabling real-time adjustments to the snowmaking operations.

[0221] The control unit 38 of Fig. 1 acts as the central hub for the system for controlling snow making machines 10, processing incoming data from the sensors of the snow making machines 40 and generating predictive analyses for snow depth across the ski slope 44. The control unit utilizes this information to issue control signals to the snow making machines 40, directing them to adjust their operations to achieve the intended snow depth in each zone 48.

[0222] A snow reservoir 52 is also depicted in FIG. 2, serving as a strategic reserve for snow that can be used to supplement the natural snow cover on the ski slope 44. The snow reservoir 52 is particularly useful during periods of high demand or when weather conditions are not conducive to snow production. Snow stored in the snow reservoir 52 can be distributed to various zones 48 on the ski slope 44 as per the requirements, ensuring consistent and high-quality snow coverage. M / TAL-012-PC

[0223] - 41 -

[0224] The system for controlling snow making machines 10 is designed to be dynamic and responsive, with the ability to adapt to changing conditions on the ski slope 44. The control unit 38, in communication with the snow making machines 40, issues control signals that not just adjust operational parameters but also guide the movement of the snow making machines 40 across the ski slope 44. This adaptability is especially beneficial for large ski resorts, where conditions can vary greatly across different areas of the slope.

[0225] It should be noted that in the context of outputting information and recommendations, the output module 30 is configured to provide a comprehensive suite of actionable insights tailored to the management of the ski slope 44. While the outputted information may include detailed snow depth charts for each virtual segment 46, highlighting certain areas where snow depth is below or above the desired thresholds. Recommendations may comprise a variety of snow management strategies, such as targeted snowmaking in segments with deficient snow levels, redistribution of snow from areas of excess to areas of scarcity, and adjustments to the operational schedule of snow grooming equipment to optimize the snow surface for upcoming events or peak visitor periods. Furthermore, the system may advise on the strategic deployment of snowmaking resources during anticipated periods of low natural snowfall, based on long-term weather forecasts integrated into the prediction model. Additionally, the output module 30 may generate alerts for potential safety hazards, such as the risk of avalanche in segments where snow accumulation exceeds safe limits, and suggest preventive measures such as controlled detonations or temporary slope closures. These outputs are designed to support ski slope operators in making informed, data-driven decisions that enhance the safety, quality, and sustainability of the skiing experience.

[0226] EXAMPLES

[0227] Example 1: Predicting Snow Depth

[0228] Table 1 provides an example of the data inputs used by the machine learning algorithm to predict snow depth on a ski slope. The table includes historical snow data for the ski slope, including recorded measurements of snow depth at various M / TAL-012-PC

[0229] - 42 - locations on the slope and associated ambient conditions at the time of each measurement. The table also includes current snow data for various locations on the slope, including measurements of snow depth and associated ambient conditions. The machine learning algorithm may use this data to generate a prediction model for forecasting snow depth evolution on the ski slope.

[0230] Table 1: Historical and Current Snow Data Inputs

[0231] Example 2: Generating Control Signals

[0232] Table 2 provides an example of the control signals generated by the system based on the predicted snow depth. The table includes control signals for a plurality of snow making machines, each assigned to make snow directed to a specific zone of the ski slope. The control signals are configured to collectively adjust the snow production rate of the snow making machines based on the predicted snow depth for the associated zones of the slope.

[0233] Table 2: Control Signals for Snow Making Machines

[0234] Example 3: Outputting Information and Recommendations M / TAL-012-PC

[0235] - 43 -

[0236] Table 3 provides an example of the information and recommendations outputted by the system based on the predicted snow depth. The table includes expected snow conditions, potential areas of snow deficiency, and advisories for snowmaking. The table also includes specific actions for snow management such as initiating snowmaking operations, adjusting snowmaking parameters, or redistributing snow on the ski slope.

[0237] Table 3: Information and Recommendations Based on Predicted Snow Depth Example 4: Incorporating Weather Forecasts

[0238] Table 4 provides an example of how the system incorporates weather forecasts into the prediction model to enhance the accuracy of the snow depth forecasts. The table includes weather forecast data predicting future ambient conditions affecting the ski slope. The machine learning algorithm incorporates this weather forecast data into the prediction model to improve the predictive accuracy of future snow depth on the ski slope.

[0239] Table 4: Incorporation of Weather Forecasts into Prediction Model

[0240] Example 5: Adjusting Snow Production

[0241] Table 5 provides an example of how the system adjusts snow production based on real-time ambient conditions and feedback from sensors located at the ski slope. M / TAL-012-PC

[0242] - 44 -

[0243] The table includes real-time weather data, real-time energy consumption data, and feedback from sensors situated in the snow making machines. The system may use this data to generate control signals that adjust the operations of the snow making machines, optimizing snow production in response to changing conditions and promoting energy-efficient operation of the snow making machines.

[0244] Table 5: Adjusting Snow Production Based on Real-Time Data

[0245] The tables provide examples of data inputs, control signals, and system outputs that illustrate the functionality of the system for controlling snow making machines.

[0246] INDUSTRIAL APPLICATION

[0247] The invention is susceptible of industrial application in the field of ski resort management, particularly in the context of snow production and maintenance on ski slopes. The invention leverages machine learning techniques to predict snow depth on ski slopes and control snow making machines, thereby optimizing snow production and distribution. This can result in substantial energy and resource savings, as well as improved skiing conditions.

[0248] The invention can be used to control a fleet of snow making machines, adjusting the rate of snow production, the direction of the snow spray, or the distribution of the snow on the slope based on the predicted snow depth. This can result in a more efficient use of resources and a more precise control of snow production.

[0249] The system for controlling snow making machines is designed to provide comprehensive recommendations for ski slope management, including decisions such as closing a slope to preserve its condition for an upcoming ski race. If a ski M / TAL-012-PC

[0250] - 45 - race is scheduled, the system can recommend closing a slope in advance to ensure that the snow is undisturbed and in prime condition for the event. This recommendation would be outputted by the output module, which communicates such advisories to the ski slope operators through the user interface. These recommendations are based on a variety of factors, including the predicted weather conditions, the current snowpack state, and the anticipated skier load. By considering these factors, the system for controlling snow making machines ensures that the ski slope is maintained in the desired condition, optimizing the skiing experience and preserving the slope for competitive events.

[0251] In addition to ski resorts, the invention can also be used in other settings where snow production is a concern. For example, it can be used in the film industry to create artificial snow scenes, or in sports events that require artificial snow, such as snowboarding or cross-country skiing competitions.

[0252] Furthermore, the invention can be used for road maintenance and safety, particularly in regions with harsh winter conditions. The machine learning algorithm could be trained to predict ice and snow accumulation on roads based on historical data, weather forecasts, and real-time data from road sensors. This could help road maintenance crews optimize their efforts, reducing the use of salt and other de-icing materials and improving road safety.

[0253] In the field of environmental conservation, the invention can be used to monitor and manage snow conditions in sensitive ecosystems. By predicting snow depth and controlling snow production, the invention can help maintain the health of these ecosystems and reduce the impact of human activities.

[0254] The disclosed system for controlling snow making machines, while primarily designed for ski slope maintenance, can be adapted for agricultural applications in regions where snow accumulation impacts farming operations. For instance, the system could be used to predict snowmelt and its subsequent availability for crop irrigation. By accurately forecasting snow depth and melt rates using the machine learning algorithm, farmers could plan irrigation schedules more effectively, M / TAL-012-PC

[0255] - 46 - optimizing water usage and potentially reducing the reliance on other water sources during the thawing season.

[0256] Additionally, the system could be employed to manage snow cover in orchards or vineyards, where a consistent snow layer acts as insulation to protect roots and dormant plants from freezing temperatures. By controlling snowmaking machines, the system could ensure adequate snow coverage to safeguard crops through harsh winter conditions.

[0257] Furthermore, the predictive capabilities of the system could be utilized to anticipate and mitigate the risk of snow-related damages to greenhouses and other agricultural infrastructure. By predicting excessive snow accumulation, the system could trigger alerts for preemptive snow removal or reinforcement of structures, thus preventing potential collapses and loss of crops.

[0258] Finally, the invention can be used in the field of climate research. By collecting and analyzing data on snow depth and other related variables, the invention can contribute to our understanding of snow dynamics and their response to climate change. This can support the development of more accurate climate models and inform policy decisions related to climate change mitigation and adaptation.

[0259] In a structured and itemized manner, the invention can be characterized by the following itemized list of features:

[0260] 1. A computer-implemented method for predicting snow depth on a ski slope, the method comprising: receiving historical snow data related to snow depth on the ski slope, wherein the historical snow data includes recorded measurements of snow depth at various locations on the ski slope and associated ambient conditions at the time of each measurement; training a machine learning algorithm with the historical snow data; generating, by the trained machine learning algorithm, a prediction model for forecasting snow depth evolution on the ski slope; receiving current snow data for various locations on the ski slope including measurements of snow depth and associated ambient conditions; M / TAL-012-PC

[0261] - 47 - applying the current snow data; predicting snow depth on various locations on the ski slope using the prediction model; and outputting information and / or recommendations based on the predicted snow depth.

[0262] 2. The method of item 1, wherein the ski slope is divided into a plurality of virtual segments, each virtual segment representing a grid portion of the ski slope for which snow depth is predicted individually, thereby enabling localized snow management and precision in snow depth forecasting.

[0263] 3. The method of item 1 or 2, wherein the ambient conditions include at least temperature, and optionally additional weather data such as hours of sunshine, wind speed and direction, and humidity.

[0264] 4. The method of one of items 1 to 3, further comprising receiving a weather forecast including predictions of future ambient conditions affecting the ski slope; and incorporating the weather forecast into the prediction model to enhance the accuracy of snow depth forecasts.

[0265] 5. The method according to one of items 1 to 4, wherein the outputted information includes at least one of expected snow conditions, potential areas of snow deficiency, and advisories for snowmaking; and wherein the recommendations include at least one of specific actions for snow management such as initiating snowmaking operations, adjusting snowmaking parameters, or redistributing snow on the ski slope.

[0266] 6. The method according to one of items 1 to 5, further comprising receiving data related to slope-specific factors that do not change over time, and wherein the machine learning algorithm is further trained based on the slope-specific factors, thereby enabling the machine learning algorithm to extrapolate and output information and predictions for slopes with limited or no historical snow data by leveraging learned patterns from slopes with comprehensive historical data. M / TAL-012-PC

[0267] - 48 -

[0268] 7. The method according to one of items 1 to 6, further comprising generating at least one of a) optimistic and pessimistic predictions of snow depth for zones of the ski slope, wherein the optimistic predictions assume favorable weather conditions for snow retention and the pessimistic predictions assume adverse weather conditions, thereby providing a range of potential snow depth outcomes, and b) short-term and long-term predictions of snow depth for zones of the ski slope based on the trained machine learning algorithm, wherein each zone comprises a group of virtual segments.

[0269] 8. The method according to one of items 1 to 7, wherein the machine learning algorithm is configured to apply the prediction model to a plurality of ski slopes, each with potentially different slope-specific factors, to predict snow depth across multiple slopes, thereby enabling a comprehensive snow management strategy for a ski resort with diverse terrain.

[0270] 9. The method according to one of items 1 to 8, further comprising adjusting the predictions of snow depth based on real-time feedback from snow making machines deployed on the ski slope, wherein the feedback includes data on the efficacy of snow production by the snow making machines, thereby refining the prediction model for increased accuracy in subsequent forecasts.

[0271] 10. A computer-implemented method for controlling snow making machines (40) on a ski slope (44), the method comprising: predicting snow depth on the ski slope using the method according to one of items 1 to 9; generating control signals for the snow making machines based on the predicted snow depth; and controlling the snow making machines (40) based on the control signals.

[0272] 11. The method of item 10, further comprising generating control signals for a plurality of snow making machines (40), each assigned to make snow directed to a specific zone of the ski slope corresponding to a group of virtual segments (48), M / TAL-012-PC

[0273] - 49 - wherein the control signals are configured to collectively adjust the snow production rate of the snow making machines based on the predicted snow depth for the associated zones of the slope.

[0274] 12. The method of item 10 or 11, further comprising generating control signals to direct a subset of the snow making machines to produce snow for storage in a snow reservoir (52), wherein the snow produced for the reservoir is intended for later distribution onto the ski slope based on the predicted snow depth and the operational requirements of the ski slope.

[0275] 13. The method according to one of items 10 to 12, wherein the control signals are further configured to adjust the operations of the snow making machines based on real-time ambient conditions and feedback from sensors located at the ski slope, particularly sensors situated in the snow making machines, thereby optimizing snow production in response to changing weather patterns and ensuring that the snow depth within the group of virtual segments remains within a predetermined range.

[0276] 14. The method according to one of items 10 to 13, wherein the control signals are adapted to initiate or cease snow production by the snow making machines based on a threshold snow depth value determined for the group of virtual segments, and to modulate the mixture of water and air in the snow making machines to produce snow of a desired quality and consistency for the group of segments.

[0277] 15. The method according to one of items 10 to 14, wherein the control signals are generated in consideration of energy consumption, thereby promoting energyefficient operation of the snow making machines across the group of segments.

[0278] 16. A system for predicting snow depth on a ski slope, the system comprising: a data receiver module (22) configured to receive historical snow data related to the ski slope, including recorded measurements of snow depth at various locations on the ski slope and associated ambient conditions at the time of each measurement; M / TAL-012-PC

[0279] - 50 - a machine learning module (24) configured to train a machine learning algorithm with the historical snow data and to generate a prediction model for forecasting snow depth evolution on the ski slope; the data receiver module (22) further configured to receive current snow data and current ambient condition data for the ski slope; an application module (28) configured to apply the current snow data and the current ambient condition data to the prediction model; a prediction module (26) configured to predict snow depth on the ski slope using the prediction model; and an output module (32) configured to output information and / or recommendations based on the predicted snow depth.

[0280] 17. The system of item 16, wherein the data receiver module (22) is further configured to receive data related to slope-specific factors including at least one of slope gradient, aspect, and shading patterns, which are used by the machine learning module (24) to enhance the accuracy of the prediction model.

[0281] 18. The system of item 16 or 17, wherein the output module (32) is configured to provide a graphical representation of the predicted snow depth over time for various locations on the ski slope, thereby enabling visual comparison of current predictions with historical snow data.

[0282] 19. The system according to one of items 16 to 18, wherein the machine learning module (24) is further configured to update the prediction model dynamically based on feedback received from the snow making machines (40) regarding the actual snow depth after snow making operations, thereby improving the predictive accuracy of the system over time.

[0283] 20. The system according to one of items 16 to 19, wherein the data receiver module (22) is further configured to receive weather forecast data predicting future ambient conditions, and the machine learning module (24) is configured to incorporate the weather forecast data into the prediction model to improve the predictive accuracy of future snow depth on the ski slope. M / TAL-012-PC

[0284] - 51 -

[0285] 21. A system for controlling snow making machines (40) on a ski slope (44), the system comprising: a system for predicting snow depth on a ski slope according to one of items 16 to 20; a control module (30) communicatively coupled to the snow making machines (40) and configured to generate control signals based on the predicted snow depth; and the snow making machines (40) configured to receive the control signals from the control module (30) and adjust snow production in response to the control signals.

[0286] 22. The system according to item 21, wherein the control module (30) is further configured to generate control signals for a plurality of snow making machines (40) assigned to specific zones of the ski slope corresponding to a group of virtual segments (48) based on the predicted snow depth, wherein the control signals are configured to collectively adjust the snow production rate of the snow making machines (40) for the associated zones of the slope, and wherein the zones can be managed independently or collectively, allowing for zone-specific or comprehensive snow management.

[0287] 23. The system according to item 21 or 22, wherein the control module (30) is further configured to generate control signals for a plurality of snow making machines (40) that are mobile and can be moved to different zones on the ski slope based on the predicted snow depth, thereby providing flexibility in snow production.

[0288] 24. The system according to one of items 21 to 23, wherein the control module (30) is further configured to generate control signals for the snow making machines (40) based on real-time weather data, real-time energy consumption data, thereby allowing for dynamic and adaptive snow production in response to changing conditions and promoting energy-efficient operation of the snow making machines.

[0289] 25. The system according to any one of items 21 to 24, wherein the control module (30) is further configured to generate control signals to direct a subset of the snow making machines (40) to produce snow for storage in a snow reservoir M / TAL-012-PC

[0290] - 52 -

[0291] (52), wherein the snow produced for the reservoir is intended for later distribution onto the ski slope (44) based on the predicted snow depth and the operational requirements of the ski slope. 26. A snow management system, comprising: a plurality of snow making machines (40); and a control system according to one of items 21 to 25.

[0292] 27. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform the method for predicting snow depth according to one of items 1 to 9 and / or the method of controlling snow making machines according to one of items 10 to 15. LIST OF REFERENCE NUMBERS

[0293] 10 system for controlling snow making machines

[0294] 12 data processing unit

[0295] 14 data storage unit

[0296] 16 environmental sensors

[0297] 18 external data sources

[0298] 20 internet connection

[0299] 22 data receiving module

[0300] 24 machine learning module

[0301] 26 application module

[0302] 28 prediction module

[0303] 30 output module

[0304] 32 control module

[0305] 34 output interface

[0306] 36 user interface

[0307] 38 control unit

[0308] 40 snow making machines

[0309] 42 sensors on snow making machine

[0310] 44 ski slope M / TAL-012-PC

[0311] - 53 -

[0312] 46 virtual segment

[0313] 48 zone on ski slope

[0314] 50 snow making area

[0315] 52 snow reservoir

Claims

M / TAL-012-PC- 54 -CLAIMS1. A system (10) for controlling snow making machines (40) on a ski slope (44), the system (10) comprising: a system for predicting snow depth on a ski slope (44) comprising: a data receiver module (22) configured to receive historical snow data related to the ski slope (44), including recorded measurements of snow depth at various locations on the ski slope (44) and associated ambient conditions at the time of each measurement; a machine learning module (24) configured to train a machine learning algorithm with the historical snow data and to generate a prediction model for forecasting snow depth evolution on the ski slope (44); the data receiver module (22) further configured to receive current snow data and current ambient condition data for the ski slope (44); an application module (26) configured to apply the current snow data and the current ambient condition data to the prediction model; a prediction module (28) configured to predict snow depth on the ski slope (44) using the prediction model; and a control module (32) configured to generate, based on the predicted snow depth, at least one of a) control signals for automatic transmission to the snow making machines (40) to control at least one of starting snow production, stopping snow production, adjusting snow goals like specified snow amounts, adjusting water flow rates, and adjusting air pressure, and b) control information for presentation on a user interface (36) for manual control of the snow making machines (40).

2. The system of claim 1, wherein the data receiver module (22) is further configured to receive real-time feedback regarding actual snow data, and wherein the machine learning module (24) is configured to continuously update the prediction model based on the real-time feedback and / or wherein the prediction module (28) is configured to continuously update the prediction model based on the real-time feedback.M / TAL-012-PC- 55 -3. The system (10) of claim 2, wherein the real-time feedback is provided by sensors deployed on at least one of snow making machines (42), snow groomers, drones, satellites, lift systems, and fixed measurement points on the ski slope (44).

4. The system (10) of claim 3, wherein the sensors (16) comprise at least one of Light Detection and Ranging (LIDAR) devices, thermal sensors, ultrasonic sensors, ground-penetrating radar, and weather sensors for measuring snow conditions, and optionally GPS units for determining sensor location.

5. The system (10) of one of claims 1 to 4, wherein the data receiver module (22) is further configured to receive at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slope-specific data related to slope-specific factors, said slope-specific factors including at least one of slope gradient, aspect, and shading patterns, and the machine learning module (24) is configured to incorporate the at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slope-specific data into the prediction model.

6. The system (10) of one of claims 1 to 5, wherein the ski slope (44) is divided into a plurality of virtual segments (46), each virtual segment (46) representing a grid portion of the ski slope (44) for which snow depth is predicted individually, and wherein the control module (32) is configured to generate individual control signals for each virtual segment (46).

7. The system (10) of claim 6, wherein the control module (32) is configured to generate control signals for a plurality of snow making machines (40) assigned to specific zones (48) of the ski slope (44) corresponding to a group of virtual segments (46) based on the predicted snow depth, wherein the control signals are configured to collectively adjust the snow production rate of the snow making machines (40) and / or adjust snow goals like specified snow amounts for the associated zones (48) of the ski slope (44), and wherein the zones (48) can be managed independently or collectively, allowing for zone-specific or comprehensive snow management.M / TAL-012-PC- 56 -8. The system (10) of one of claims 1 to 7, wherein the control module (32) is configured to generate control signals for a plurality of snow making machines (40) that are mobile and can be moved to different zones (48) on the ski slope (44) based on the predicted snow depth.

9. The system (10) of one of claims 1 to 8, wherein the control module (32) is configured to generate control signals for the snow making machines (40) based on real-time weather data and real-time energy consumption data to adjust snow production under energy efficiency aspects.

10. The system (10) of one of claims 1 to 9, wherein the control module (32) is configured to generate control signals to direct a subset of the snow making machines (40) to produce snow for storage in a snow reservoir (52), wherein the snow produced for the snow reservoir (52) is intended for later distribution onto the ski slope (44) based on the predicted snow depth and operational requirements of the ski slope (44).

11. A method for controlling snow making machines on a ski slope, the method comprising: predicting snow depth on the ski slope using a method for predicting snow depth on a ski slope comprising: receiving historical snow data related to snow depth on the ski slope, wherein the historical snow data includes recorded measurements of snow depth at various locations on the ski slope and associated ambient conditions at the time of each measurement; training a machine learning algorithm with the historical snow data; generating, by the trained machine learning algorithm, a prediction model for forecasting snow depth evolution on the ski slope; receiving current snow data for various locations on the ski slope including measurements of snow depth and associated ambient conditions; applying the current snow data and the current ambient condition data to the prediction model;M / TAL-012-PC- 57 - predicting snow depth on various locations on the ski slope using the prediction model; and generating, based on the predicted snow depth, at least one of a) control signals for automatic transmission to the snow making machines to control at least one of starting snow production, stopping snow production, adjusting snow goals like specified snow amounts, adjusting water flow rates, and adjusting air pressure, and b) control information for presentation on a user interface for manual control of the snow making machines.

12. The method of claim 11, further comprising receiving real-time feedback regarding actual snow data and continuously updating the prediction model and / or the snow depth prediction based on the real-time feedback.

13. The method of one of claims 11 or 12, further comprising receiving at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slopespecific data related to slope-specific factors, said slope-specific factors including at least one of slope gradient, aspect, and shading patterns, and incorporating the at least one of weather forecast data, real-time weather data, data from lift systems, data from ticketing and access control systems, data from power plants, and slopespecific data into the prediction model.

14. The method of one of claims 11 to 13, wherein the ski slope is divided into a plurality of virtual segments representing a grid portion of the ski slope for which snow depth is predicted individually, and wherein the method comprises generating individual control signals for each virtual segment.

15. The method of claim 14, comprising generating control signals for a plurality of snow making machines assigned to specific zones of the ski slope corresponding to a group of virtual segments based on the predicted snow depth, wherein the control signals collectively adjust the snow production rate of the snow making machines and / or adjust snow goals like specified snow amounts for the associated zones of the ski slope, and wherein the zones are managed independently or collectively for zone-specific or comprehensive snow management.M / TAL-012-PC- 58 -16. The method of one of claims 11 to 15, comprising generating control signals for a plurality of snow making machines that are mobile and repositioning the mobile snow making machines to different zones on the ski slope based on the predicted snow depth.

17. The method of one of claims 11 to 16, comprising generating control signals for the snow making machines based on real-time weather data and real-time energy consumption data to adjust snow production under energy efficiency aspects.

18. The method of one of claims 11 to 17, comprising generating control signals to direct a subset of the snow making machines to produce snow for storage in a snow reservoir, wherein the snow produced for the snow reservoir is intended for later distribution onto the ski slope based on the predicted snow depth and operational requirements of the ski slope.

19. The method of one of claims 11 to 18, further comprising at least one of the following features: wherein the ambient conditions include at least temperature, and optionally additional weather data such as hours of sunshine, wind speed and direction, and humidity; outputting information on at least one of expected snow conditions, potential areas of snow deficiency, and advisories for snowmaking; outputting recommendation related to at least one of specific actions for snow management such as initiating snowmaking operations, adjusting snowmaking parameters, or redistributing snow on the ski slope; receiving data related to slope-specific factors that do not change over time, and wherein the machine learning algorithm is further trained based on the slopespecific factors, thereby enabling the machine learning algorithm to extrapolate and output information and predictions for slopes with limited or no historical snow data by leveraging learned patterns from slopes with comprehensive historical data; wherein the machine learning algorithm is configured to apply the prediction model to a plurality of ski slopes, each with potentially different slope-specificM / TAL-012-PC- 59 - factors, to predict snow depth across multiple slopes, thereby enabling a comprehensive snow management strategy for a ski resort with diverse terrain.

20. The method of one of claims 11 to 19, further comprising generating at least one of a) optimistic and pessimistic predictions of snow depth for zones of the ski slope, wherein the optimistic predictions assume favorable weather conditions for snow retention and the pessimistic predictions assume adverse weather conditions, thereby providing a range of potential snow depth outcomes, and b) short-term and long-term predictions of snow depth for zones of the ski slope based on the trained machine learning algorithm.

Images